Citation
Development o a standard of testing and evaluation for 3D-printed pediatric upper limb prosthetics

Material Information

Title:
Development o a standard of testing and evaluation for 3D-printed pediatric upper limb prosthetics
Creator:
Lyle, Brendan Robert ( author )
Place of Publication:
Denver, Colo.
Publisher:
University of Colorado Denver
Publication Date:
Language:
English
Physical Description:
1 electronic file (86 pages) : ;

Thesis/Dissertation Information

Degree:
Master's ( Master of science)
Degree Grantor:
University of Colorado Denver
Degree Divisions:
Department of Bioengineering, CU Denver
Degree Disciplines:
Bioengineering

Subjects

Subjects / Keywords:
Prosthesis -- Design and construction ( lcsh )
Three-dimensional printing ( lcsh )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Review:
The purpose of this study was to identify which common polymers utilized for 3D-printing provide optimum mechanical properties for use in pediatric upper limb prosthetics and to develop a standard method of failure mode testing which can be applied across a wide range of devices. Selected devices were also tested for their ability to complete existing standard usability tasks. Results of mechanical testing showed that ABS filament provided the most robust material properties, and that print orientation showed no significant effect on ultimate tensile strength. The standard failure mode test developed showed consistent results across all designs tested and was quickly repeatable and accessible to those unfamiliar with mechanical testing. Tasks requiring a tripod or lateral prehensile grip on lightweight objects were most successful, while large and heavy objects and spherical prehensile grip were the least. These findings will be useful in the development of a standard of testing and evaluation for 3D-printed upper limb prosthetics so that they can be better understood by the medical and clinical community, identifying their strengths and weaknesses in a straightforward manner.
Bibliography:
Includes bibliographical references.
System Details:
System requirements: Adobe Reader.
Statement of Responsibility:
by Brendan Robert Lyle.

Record Information

Source Institution:
University of Colorado Denver
Holding Location:
Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
on10230 ( NOTIS )
1023030238 ( OCLC )
on1023030238
Classification:
LD1193.E56 2017m L95 ( lcc )

Downloads

This item has the following downloads:


Full Text
DEVELOPMENT OF A STANDARD OF TESTING AND EVALUATION FOR
3D-PRINTED PEDIATRIC UPPER LIMB PROSTHETICS
by
BRENDAN ROBERT LYLE
B.S., Colorado School of Mines, 2014
A thesis submitted to the Faculty of the Graduate School of the University of Colorado in fulfillment of the requirements for the degree of Master of Science Bioengineering Program
2017


This thesis for the Master of Science degree by Brendan Robert Lyle has been approved for the Bioengineering Program by
Richard Weir, Chair Cathy Bodine Steven Lammers
Date: December 16, 2017


Lyle, Brendan R. (M.S., Bioengineering Program)
Development of a Standard of Testing and Evaluation for 3D-Printed Pediatric Upper Limb Prosthetics
Thesis directed by Research Associate Professor Richard Weir
ABSTRACT
The purpose of this study was to identify which common polymers utilized for 3D-printing provide optimum mechanical properties for use in pediatric upper limb prosthetics and to develop a standard method of failure mode testing which can be applied across a wide range of devices. Selected devices were also tested for their ability to complete existing standard usability tasks. Results of mechanical testing showed that ABS filament provided the most robust material properties, and that print orientation showed no significant effect on ultimate tensile strength. The standard failure mode test developed showed consistent results across all designs tested and was quickly repeatable and accessible to those unfamiliar with mechanical testing. Tasks requiring a tripod or lateral prehensile grip on lightweight objects were most successful, while large and heavy objects and spherical prehensile grip were the least. These findings will be useful in the development of a standard of testing and evaluation for 3D-printed upper limb prosthetics so that they can be better understood by the medical and clinical community, identifying their strengths and weaknesses in a straightforward manner.
The form and content of this abstract are approved. I recommend its publication
Approved: Richard Weir.


ACKNOWLEDGEMENTS
The author would like to acknowledge the following individuals for their contributions and assistance in the completion of this study: Dr. Richard Weir Dr. Cathy Bodine Dr. Steven Lammers
Dr. Levin Sliker
Dr. Bradford Smith
Dr. Travis Heare and the Amputee Program at Childrens Hospital Colorado
Stephen Huddle


TABLE OF CONTENTS
I. INTRODUCTION.................................................................1
LITERATURE REVIEW.............................................................2
Part 1: Background on Childhood Amputations and Pediatric Prosthetics.......2
Amputation Causes.........................................................2
Pediatric Amputee Statistics..............................................6
Age for Fitting and Outcome Effect........................................7
Prosthetic Requirements and Types.........................................9
Outcomes of Prosthetic Use and Design Evaluation.........................15
Part 2: 3D printing and the Current State of 3D-Printed Prosthetics........20
3D Printing Hardware.....................................................20
FDM Plastics.............................................................24
3D Printed Prosthetics...................................................25
SPECIFIC AIMS................................................................27
II. EXPERIMENT AND RESULTS.....................................................28
Experimental Set-up........................................................28
Construction of 3D-printed hands.........................................28
Inclusion and Exclusion Criteria for 3D-Printed Hand Designs.............29
Printers and Software....................................................32
Selection of Plastics....................................................34
Tensile Testing..........................................................35
Lateral Testing for Standard Development.................................38
Usability Modified SHAP test...........................................42
Results....................................................................44
Tensile Testing..........................................................44
Lateral Failure Mode Testing.............................................48
SHAP Task Results........................................................57
Discussion.................................................................59
Construction Observations................................................59
Tensile Test Results.....................................................65
Lateral Failure Mode Testing.............................................67
Modified SHAP Test.......................................................70
v


Conclusion...................................................72
REFERENCES.......................................................75
VI


LIST OF TABLES
Table 1 Student t-test p-values of tensile tested plastics. Upper left Comparing plastic types in Flat orientation. Lower left Comparing plastic types in Vertical print orientation. Right Comparing Flat vs Vertical Print orientations for each tested plastic.
......................................................................................48
Table 2 SHAP Task results from each tested hand on pass/fail basis and primary
prehensile pattern utilized for each task.............................................58
Table 3 Overall Completion rates from each hand design based on primary prehensile
pattern
59


LIST OF FIGURES
Figure
1 - Prosthetics for child effected by thalidomide Science Museum of London, licensed
under CC BY-SA 2.0.................................................................4
2 - Hand of infant affected by Amniotic Band Syndrome (Image in public domain)....5
3 - Time to failure of attending patients based on age of initial fitting (Davids et al. 2006,
pg.1296). Used with permissions by Wolters Kluwer Health, Inc......................9
4 - Passive mitten (Krebs et. al, 1991) Used with permissions by Oxford University Press 10
5 - Child sized voluntary-opening hook (Krebs et al. 1991, pg.927). Used with
permissions by Oxford University Press............................................11
6 - Child's myoelectric hand and glove (Krebs et al. 1991, pg.926). Used with
permissions by Oxford University Press............................................13
7- a) Raptor Hand Render by Jeremy Simon is licensed under CC-BY 3.0 b) Raptor Reloaded c) Flexy Hand d) Cyborg Beast Render by Creighton Labs is licensed under CC BY-NC.............................................................................29
8 - Phoenix Hand v2 by EnableCommunityFoundation is licensed under CC BY-NC ... 31
9 - Failed Pheonix Hand v2 print due to insufficient surface area for proper filament
adhesion..........................................................................32
10 - ASTM D638 Type 1 Tensile Specimen W=13mm; L=57mm; WO=19mm;
LO=165mm; G=50mm; D=115mm; R=76mm; T=3.2mm. Wc and WO are defined in ASTM D638 for alternate specimen types or specific materials not utilized in this study (ASTM International, 2014)........................................................36
11 - Carbon-Fiber (left) and Reinforced Nylon (right) Type 1 Tensile Specimens...36
viii


12 - a) Flat print orientation b) Vertical print orientation.......................37
13 - a) Raptor test jig b) Raptor Reloaded test jib c) Cyborg Beast test jig d) Flexy Hand
test jig............................................................................39
14 - a) Test jig mounting bracket b) Lateral Failure Mode tensile hook.............40
15 Laterail Failure Mode Test Setup (post break)..................................41
16 Raptor Hand with Lee Tippi Micro Gel fingertips................................43
17 - Camry EH101 hand dynamometer................................................43
18 - Stress Strain curves of the Flat and Vertical Generic ABS tensile specimens...44
19 - Stress Strain curves of Flat and Vertical Hatchbox ABS tensile specimens.......45
20 - Stress Strain curves of Flat and Vertical RGD525 tensile specimens.............45
21 - Mean Ultimate Tensile Strengths from all specimens and print orientations with
standard error.......................................................................46
22 - Mean Ultimate Tensile Strengths from all specimens and print orientations with
expected literature values and composite samples.....................................47
23 - Cyborg Beast Lateral Failure Mode Test results................................49
24 - Flexy Hand Lateral Failure Mode Test results..................................49
25 - Raptor Lateral Failure Mode Test results......................................50
26 - Raptor Reloaded Lateral Failure Mode Test results.............................50
27 - Mean Failure Strengths of the tested hand designs versus strain rate with standard
deviation............................................................................51
28 - Raptor ANOVA Table.............................................................52
29 - Raptor ANOVA table statistics..................................................52
30 - Raptor Tukey HSD results.......................................................53
ix


31 - Raptor Reloaded ANOVA Table......................................................53
32 - Raptor Reloaded ANOVA Table statistics..........................................54
33 - Raptor Reloaded Tukey HSD results...............................................54
34 - Cyborg Beast ANOVA table........................................................55
35 - Cyborg Beast ANOVA table statistics.............................................55
36 - Cyborg Beast Tukey HSD results..................................................56
37 - Flexy Hand ANOVA table...........................................................56
38 - Flexy Hand ANOVA table statistics................................................57
39 - Flexy Hand Tukey HSD results....................................................57
40 - Render of Flexy Hand finger build plate.........................................62
41 - Print failure of Flexy Hand printer build plate due to improperly constructed CAD
file..................................................................................63
42 - Example of tension pin splitting common among all designs tested................64
43 - Gauntlet failures of Flexy Hand (a) and Raptor (b). Also note the tension pin block
split in the Flexy Hand gauntlet from tightening of flexion lines.....................64
44 - Proximal Phalange joint failure experienced in all lateral mode failure tests. From
left to right: Cyborg Beast, Raptor Reloaded, Raptor..................................69


I. INTRODUCTION
The rise of 3D-printing technology in the last decade has caused a surge of interest in a wide array of fields of medicine. The use of 3D-printed materials for custom-made, inexpensive, and quickly produced prosthetic devices has gained worldwide attention through media and popular culture (Murphy, 2017). While these reports provide some insight into the design of hands and how they may be used, there is a distinct lack of information into their benefits over existing devices. Reviews of the technology and apparent benefits exist, but they tend to make many generalizations about the capability of 3D printed devices, while not providing first-hand validation (Kate,
Smit, & Breedveld, 2017). What is not being addressed is the lack of understanding among clinicians, occupational therapists, patients, and their families as to how these devices should best be used or what considerations need to be made for the devices to be successful and improve the quality of life for the user.
Designs for 3D-Printed Prosthetic hands are available online from many sources, but e-NABLE is perhaps the most well-known, providing resources for creators, makers, health professionals, and patients to connect and receive help in getting a device printed to meet their needs (e-NABLE, 2017). E-NABLE has provided designs used in previous studies of 3D-printed prosthetics, but these are generally literature reviews of available designs, their basic functions, and recommendations for fitting and sizing (Zuniga, et al., 2015). Many studies done with traditional prosthetics show that abandonment rates of users are commonly in excess of 50% if the device does not meet their needs or fails to improve their functionality, and these rates climb even higher for children (Biddiss & Chau, 2007). What the literature lacks is an engineering analysis 3D-printed prosthetic
1


designs, and effective guidelines written with clinicians and pediatric prosthetic specialists in mind for what considerations must be made for fabricating prosthetics using 3D-printers and how best they can be utilized.
This project will address these issues and provide the basis on which production standards and future evaluations can be done and answer the question of what do production standards for 3D-printed prosthetics need to address, and how should they be conducted. The results of these tests and examinations can be utilized in the future in academic and industry publications to formulate standards as well as guidelines to make inexpensive, functional, and appealing prosthetics a viable choice for childhood amputees and their families, as well as provide an analysis of existing designs and 3D printing technology to further improve the development cycle of new devices moving forward.
LITERATURE REVIEW
Part 1: Background on Childhood Amputations and Pediatric Prosthetics
Amputation Causes. Knowing the cause of an amputation and understanding how they are categorized is important in establishing what treatment options individual patients should receive, and how their needs should be addressed. There are two classifications under which all amputations fall, congenital limb deficiency and acquired amputations. Congenital limb deficiency is a broad term that encompasses any birth defect that effect the childs arms or legs. In the US, this is present in less than 1 out of every 1,000 births. Deficiencies are more likely to occur in upper limbs than in lower limbs, and even more unlikely for a deficiency to be present in both the lower or upper extremities. Total congenital limb loss is also rare (Smith, 2006). Smith referred to all
2


other deficiencies as acquired limb deficiency, encompassing loss of limb ranging from vascular disorder, cancer, traumatic injury, etc.
Smith wrote from the perspective of giving information to the parents of children born with congenital amputations, who are largely responsible for deciding what care their child receives and what treatments and prosthetic devices they receive. They are also responsible for following up with treatments provided by clinicians and occupational therapists. Smiths brief overview in inMotion, an online publication of the Amputee Coalition provided background information on congenital amputations that is easily accessible to the parents of childhood amputees to familiarize them with their childs condition. The concerns of the parents of childhood amputees are a key aspect to recognize in the development of prosthetics as well, as they ultimately must assist their child in learning how to use and adapt to their devices.
Congenital limb deficiencies are caused by a variety of factors. Exterior causes may come from exposure of the mother to hazardous chemical substances, such as the infamous outcome of thalidomide prescription to pregnant mothers in the 1950s (Krebs, Edelstein, & Thomby, 1991). The example of a disabled by the effects of thalidomide in
3


Figure 1 shows the extreme severity of limb deficiencies that occurred due to the drugs administration, prosthetic harness designed for a child
Figure 1 Prosthetics for child effected by thalidomide Science Museum of London,
licensed under CC BY-SA 2.0.
Another exterior cause is early amnion rupture, which can result in amniotic band syndrome (Davids, Wagner, Meyer, & Blackhurst, 2006). Amniotic Band Syndrome occurs when the limbs of the developing fetus are entrapped by stands of amniotic tissues, which inhibit development and cause a wide array of deletions and deformations
4


(Shetty, Menezes, Tauro, & Diddigi, 2013). An example of an infants hand effected by Amniotic Band Syndrome is shown in Figure 2.
Figure 2 Hand of infant affected by Amniotic Band Syndrome (Image in public domain).
In most cases, however, congenital limb deficiency is an isolated condition not
associated with the musculoskeletal system. These deficiencies are likely the result of mutations or vascular compromise to the apical ectodermal ridge (Davids, Wagner, Meyer, & Blackhurst, 2006). The exact cause of these mutations is unknown.
Both Krebs et al. and Davids el al. discussed the background of congenital amputations in their publications on childhood prosthetics, despite their difference in subject matter. Krebs et al. cited the use of thalidomide in prompting the growth of interest in the field of pediatric prosthetics, which at the time had been the most studied cause of childhood amputations. Davids et al. published their long-term outcome study 15 years after Krebs et al. published their overview of pediatric prosthetic devices, and
5


there is notably a shift in focus towards genetic mutation and idiopathic occurrences. The increased knowledge in the field of genetics between the publication of these two papers shows the increased understanding that researchers have gained in addressing what causes of childhood amputations are most significant. Shetty et al. (2013) further expands on the notion established by Davids et al. that Amniotic Band Syndrome can be caused by early amnion rupture, again focusing in on greater detail with a specific case study to better understand and establish the causes of congenital limb deficiencies.
Pediatric Amputee Statistics. Due to the wide range of amputation causes, types, and methods of reporting by health facilities, there is little comprehensive data on the population of childhood amputees in the United States (Krebs, Edelstein, & Thomby, 1991). Krebs et al. provides insight into the epidemiologic factors that lead to amputations in children, and sites his own study into the demographics of childhood amputees. The study from 1984 was taken by 4,105 children being treated at specialty clinics for limb deficiencies. From the data received by Krebs from these clinics, it was estimated that of all cases, 17% had been first identified in the past year, with a further three fifths being due to congenital limb reductions (Krebs, Edelstein, & Thornby, 1991). This concluded that there are 5,525 new cases of childhood amputations to be expected each year, with 3,315 of these being congenital. Citing this same study, Davids et al. stated that unilateral congenital below-elbow deficiencies are the most common limb-deficiency managed in North America, thus the need for prosthetic devices that address these deficiencies is the greatest (Davids, Wagner, Meyer, & Blackhurst, 2006).
Both Davids et al. and Krebs et al. noted the lack of available data from clinical sources on childhood amputees. Both Krebs et al. and Davids et al. cited a separate study
6


done by Krebs in 1984 on the etiology of childhood amputees, making it one of the few definitive sources on patient statistics, despite its age. However, given the endemic nature of childhood amputations, it is safe to assume that the instance rate of childhood amputees in the US has not changed significantly. Both groups also recognized from Krebs study that the most abundant group of childhood amputees suffer from unilateral congenital below-elbow deficiencies. It follows that demand for devices that address patients with these deficiencies is the highest.
Age for Fitting and Outcome Effect. The age at which a child is fitted for a prosthetic is an important consideration for clinicians. Krebs et al. identified the need to fit a prosthetic as early as possible, with ages of less than 3 being the most widely reported (Krebs, Edelstein, & Thornby, 1991). Davids et al. performed a long-term study from 1954 to 2004 with 260 childhood patients exhibiting upper-limb below-elbow deficiencies. The goal of this study was to determine the outcomes of long term prosthetic management in children with unilateral congenital below-elbow deficiency (Davids, Wagner, Meyer, & Blackhurst, 2006). A successful outcome was defined as a child and parents who continued attending the limb-deficiency clinic and claimed during follow-ups that a prosthetic had been worn outside the clinical setting for any period of time (Davids, Wagner, Meyer, & Blackhurst, 2006). An unsuccessful outcome was defined as a child and parent who no longer followed up at the clinic or reported that the child did not wear a prosthesis (Davids, Wagner, Meyer, & Blackhurst, 2006). The study found that fitting a child below the age of 3 provided the best long-term outcome for long term prosthetic use, but also recognized that fitting before the age of one had little impact on the outcome of long term prosthetic use. The most important factor in a successful
7


long-term outcome for prosthetic use was intensive training under the direction of the clinics occupational therapists, and utilizing a variety of prosthetic designs over the childs growth. This progression of designs began with a passive device, fitted once the child had learned to walk. Providing positive results and interest from the child and parents, the child was then moved into a body-powered device between the ages of two to four, and finally myoelectric devices of increasing complexity (Davids, Wagner, Meyer, & Blackhurst, 2006).
The age for fitting a child in a prosthetic device is addressed by both Davids et al. and Krebs et al. While Krebs et al. did not address a minimum age for fitting, it recognized that the optimal age for fitting is before 3 years, when the child has become old enough to understand instruction from occupational therapists in a clinical setting, and their parents outside of the clinic. Davids et al. similarly concluded that before the age of 3 gives the best outcome for long term success, with the stipulation of avoiding fitting before the age of one. Davids et aV s results of time to failure of attending patients based on age of initial fitting is shown in Figure 3.
8


1.0
c a> £ tz A < 2 o flu £
> c
t/3 C
*2 es u
S^
£ t
r; s
CX o u s ai
o
C
o
cx
o
0.9 0 8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0
"1
5 6 7 8 9 10 11 12 13 14 15
Time to Failure (Years)
Figure 3 Time to failure of attending patients based on age of initial fitting (Davids et al. 2006, pg.1296). Used with permissions by Wolters Khmer Health, Inc.
Davids et al.s study focused more on age of fitting than Krebs et al., but their findings suggest that fitting a prosthetic too early can lead to issues in the developing childs motor skills. The emphasis in both papers on this information shows the importance of fitting children with devices, rather than waiting until they are more developed. Thus, there is a clear need for prosthetics specifically designed to meet the needs of children.
Prosthetic Requirements and Types. Pediatric prosthetics require special considerations to meet the needs of children and their parents. Not only must the prosthetics be smaller, but the physical and psychological growth of the child must be considered (Krebs, Edelstein, & Thornby, 1991). The needs of the parents also factor into the prosthetic prescription; some families may want a functional device fitted on their child as soon as possible, while others may want an aesthetic passive device to disguise the amputation. (Krebs, Edelstein, & Thornby, 1991). Financial considerations must also be considered by the parents.
9


Passive devices are the simplest of all prosthetics. They consist of rubber or plastic forms modelled to look like the missing appendage in both shape and color. A passive mitten sized for a child is shown in Figure 4.
Figure 4 Passive mitten (Krebs et. al, 1991) Used with permissions by
Oxford University Press
While they are unable to provide function that replaces the deficient limb, infants fitted with passive devices have been observed using the device to stabilize their movements, such as when transitioning from sitting to standing (Krebs, Edelstein, & Thornby, 1991). However, Davids et al. suggested that a passive prosthetic should not be fitting to the child until it has learned to walk, for fear that it may interfere with developmental motor skills.
Body-driven systems are the next progression in prosthetic design. The most common and most diverse body driven terminal device is the hook. An example of a body-powered hook sized for a child is shown in Figure 5.
10


Figure 5 Child sized voluntary-opening hook (Krebs et al. 1991, pg.927). Used with
permissions by Oxford University Press.
This allows for bimanual prehension, an important motor skill in early childhood development (Krebs, Edelstein, & Thornby, 1991). The open-close motion of the hook is commonly controlled by a fitted shoulder harness, to which a cable extending from the hook is run up the arm and anchored to the opposing shoulder. Adduction of the shoulder transmits movement through the cable to open or close the attached device.
Modifications of the hook exist that more closely resemble the form and ergonomics of the human hand, providing a more natural grasping pattern and aesthetics. These devices can be covered with a cosmetic glove to further disguise the amputation. However, the gloves can discolor and tear, requiring replacement (Krebs, Edelstein, & Thornby, 1991).
The most complex prosthetics are myoelectric units. These units are driven by electronic motors to control hook or hand to open and close. Electronic Electromyogram (EMG) sensors are placed on the skin above corresponding muscles groups to detect electrical signals when the muscle is voluntarily flexed (Krausz, Rorrer, & Weir, 2015).
11


This signal corresponds to a specific motion on the device. For example, flexion of the bicep will close a hand, while flexion of the tricep will open it. This is a common arrangement, as the agonist-antagonist arrangement is easy for the user to understand and control. Depending on the level of the amputation, there may be a limited number of motor control sites available (Krausz, Rorrer, & Weir, 2015). This leaves many designs restricted to single degrees of freedom, which can be limited for adults, but in children, can provide a useful training tool. While the most functional of prosthetic devices, it is uncommon for children to be fitted with myoelectric devices. While battery powered devices eliminate the need for harnesses and cables, the unit is usually heavier and more fragile (Krebs, Edelstein, & Thornby, 1991). Additionally, the child may have to wear the battery pack on their body or waist if the space inside the proximal device and the terminal device is not large enough to house it internally. The added weight and discomfort of such devices, along with their fragility and complex operation make them less suited for children, though infants as young as 18 months have been fitted (Krebs, Edelstein, & Thornby, 1991). Like body driven hooks and hands, myoelectric devices can be fitted with aesthetic sleeves to better disguise the patients deficiency. An example of a child-sized myoelectric hand is shown in Figure 6.
12


Figure 6 Childs myoelectric hand and glove (Krebs et al. 1991, pg.926). Used with permissions by Oxford University Press.
All of these systems are classified as terminal devices, which replace a missing hand. Prosthetics also require proximal devices to mount the terminal device to the residual limb. Behind the terminal device is commonly a wrist unit, which provides passive pronation and supination. Wrist units also exist that can provide palmar flexion (Krebs, Edelstein, & Thornby, 1991). Past the wrist unit are the remaining proximal units. On body driven systems, this consists of the harness and cable system, while on myoelectric units, this will include signal processors, motor control units, and battery packs. Attachment of the proximal device to the body is done through flexible sockets with a rigid frame custom fit to the patient (Krebs, Edelstein, & Thornby, 1991). The size form of this unit is uniquely depending on the level of amputation for the patient.
As the child grows, progressively larger devices will need to be exchanged. Commercial terminal devices and wrist units typically come in a range of sizes that can be modularly swapped out (Krebs, Edelstein, & Thornby, 1991). Sockets and rigid bodies must be resized or rebuilt to fit the patients growing body.
The type of prosthetic selected for the child is dependent on many factors, as pointed out in the literature. One factor that was only briefly mentioned by Krebs et al. is
13


the factor of cost. In their study on low-cost prosthetic hands, Zuniga et al. quoted the price of a body-powered system at $4,000USD-$20,000USD (Zuniga, et al., 2015). Krausz et al. estimated the price of state-of-the-art myoelectric hands at $30,000USD, and that figure does not include the additional cost of proximal devices. Unlike adults, childhood prosthetics must be refitted and replaced to keep up in size and function as the child grows. The cumulative costs of fitting a child with such an expensive device every one to two years is high. There are many more options available for adult amputees, and for childhood amputees when funds are unlimited, but there is a distinct lack of low-cost options that provide a similar level of function. Davids et al. studied the time to abandonment in their patients based on the type of device used. Passive devices were abandoned the quickest, while myoelectric devices were abandoned the least often, with body-drive devices between. The time to failure based on the patients studied by Davids et al. who wore varies types of prosthetic devices is show in Figure 7.
Figure 7 Time to Failure of Patients based on prosthetic design (Davids et al. 2006, pg.1297). Used with permissions by Wolters Kluwer Health, Inc.
14


This shows that the use of functional prosthetics are more desirable to patients than ones that just disguise a limb deficiency.
Outcomes of Prosthetic Use and Design Evaluation. It is the goal of physical or occupational therapists working with amputees to find the best outcome for their patients, whether that includes the use of a prosthetic or no device at all. Davids et al. noted that the children in their study could perform most tasks without the use of a prosthetic, and if the device did not enhance the childs function, it would typically be abandoned. (Davids, Wagner, Meyer, & Blackhurst, 2006). Over the course of the 50-year study, 49% of Davids et al.s patients had an unsuccessful outcome. Successful outcomes were reported when the child and parents regularly attended the clinic for therapy and training in their devices (Davids, Wagner, Meyer, & Blackhurst, 2006). This suggests that the type of device used is less important than how well familiarized the patient is with it, and the level of support in its use provided. Davids et al. conceded that their study may over- or under-estimate the number of successful patient outcomes, as their pass/fail criteria was very broad.
In a similar study on device abandonment, Biddiss & Chau, looked at 25 years of prosthetic user data across 200 articles to determine what dissatisfactions users had with devices that led to their abandonment. They divided their findings based on the type of device: passive, body-powered, and myoelectric. Biddiss & Chau found that passive devices had the lowest abandonment rate, as low as 6%, but with some reports as high as 100% (Biddiss & Chau, 2007). They theorize the users relative satisfaction was due to the simplistic nature of passive devices and their cosmetic appeal. The concerns with passive devices included were wear temperature, glove malfunctions, weight, wear on
15


clothing, and irritation from straps. Rates of abandonment for body-powered devices were reported between 16% and 66%, with lower rates of rejection among adults than children (Biddiss & Chau, 2007). They recognized slowness of movement, awkward use, difficult cleaning maintenance, weight, low grip strength, and high energy expenditure as well as appearance, as the most prominent reasons for abandonment (Biddiss & Chau, 2007, p. 241). Biddiss & Chau also noted the prominent popularity of body-powered systems despite more advanced systems being available, due to their functionality, durability, lower weight, visibility of objects being handled, and accessibility of designs. Rates of abandonment for myoelectic designs ranged from 0% to 75%. Biddiss & Chau hypothesized that the low abandonment rate of myoelectric devices stem from advancing designs, availability, and the growing culture of technology in todays society. Among the same complaints of weight, appearance, and discomfort that they shared with other types of devices, myoelectric devices were reported to have the additional issues of battery life and charging, motor and sensor reliability, and an unsatisfactory range of motion in the finger and wrist units (Biddiss & Chau, 2007). While improvements on design and functionality have been prevalent in the literature, Biddiss & Chaus observations showed that there is also a need to address user comfort and reliability. Even the best device will be rejected if the user cannot use it easily.
Usability has become a prominent buzzword in the field of prosthetics and assistive technology. Dr. Linda Resnik examined usability and how it relates to the development and testing of upper-limb devices. Resnik began with the concept of device usability as part of the concept of usefulness, or the qualitative attribute that assess the
16


ease of use of device-user interfaces (Resnik, 2011, p. 698). In their work, Resnik cited the ISO definition of usability as the most relevant to medical devices:
The extent to which a product can be used by specified user to achieve specified goals with effectiveness, efficiency, and satisfaction in a specified context of use (Resnik, 2011, p. 698).
Resnik recognized that previously, there was little consensus on the most important aspects that need to be considered for evaluating the performance of prosthetics. The biggest issue in prosthetics is there are so many concerns that must be addressed in order to create a successful outcome. Resnik clearly addressed that for prosthetic usability to be improved, designs cannot be created in a vacuum. The device must meet the users needs, and that studies must be done with feedback on device performance and usability from the users perspective. This complicates research, usually resulting in mixed methods and case studies, which cannot be easily repeated and verified. It is clear from Resniks work that standards for usability, and not just device performance must be created and adopted uniformly for future developments in prosthetics to be successful.
One such system for quantifying the success in the use of a prosthetic device is the Southampton Hand Assessment Procedure (SHAP). The SHAP test measures the functionality of normal, impaired, and prosthetic hands, and is widely used by clinicians due to its thorough overview of the six prehensile grips (spherical, tripod, power, lateral, tip, extension) and a simple to understand general functionality score (Vasluian, Bongers, Reinders-Messelink, Dijkstra, & van der Sluis, 2014). The six prehensile patterns are illustrated in Figure 8.
17


Lateral
Power
/
IJ Extension
Figure 8 Six Prehensile Patterns examined by the SHAP test (Metcalf, et al., 2008) Used with permissions by SAGE Publications, Ltd.
The SHAP test involves 26 tasks, 12 where the patient manipulates abstract objects, and 14 where the patient performs an activity of daily living (ADL). The time to complete the task is recorded in seconds, and using a z-score transform of each time and the Euclidean distance, the six prehensile patterns and an overall index of function (IOF) is calculated. These scores range from 1 to 100, where 100 is normal functionality.
These scores are compared to a predetermined norm, so a score of over 100 is possible if the patient scores above the norm. These norm values are unavailable due to intellectual property rights of the commercial party that distributes the SHAP test kit (Vasluian, Bongers, Reinders-Messelink, Dijkstra, & van der Sluis, 2014).
The study by Davids et al. recognized the need for data on long-term prosthetic use, starting in early childhood. Davids et al. recognized that a 49% abandonment rate of subjects in the study is significantly high, and that one way to decrease user abandonment is to increase the variety of devices available to patients, to best suit a wide array of needs (Davids, Wagner, Meyer, & Blackhurst, 2006). The study performed by Vasluian et al. attempted to develop a new set of norm values so that the SHAP test may be applied to
18


children, in what they called the SHAP-C (Vasluian, Bongers, Reinders-Messelink, Dijkstra, & van der Sluis, 2014). Their study also modified the objects and protocols in the SHAP test itself. The size of some of the abstract and ADL objects was reduced or modified for easier manipulation by a child. Other tasks were modified so the completion criteria would be easier, such as the test assessor assisting the child in initially grasping certain objects. Both these studies showed that unique criteria must be considered not only when developing prosthetic devices for children, but also in how performance is assessed, whether it be directly assessing the device, or determining success by long-term adoption.
At the time of this writing, there are no recognized standards on evaluating the mechanical capabilities of upper-limb prosthetics. Two ISO standards exist for evaluating lower limb prosthetic. ISO 10328:2016 defines structural testing of lower-limb prostheses (ISO 10328:2016, 2017). ISO 22675:2016 defines testing for ankle-foot devices and foot units (ISO 22523:2006, 2017). Neither of these standards is publicly available, with their contents behind a paywall of $205.67USD each. This cost is a large hurdle for an independent prosthetic designer or researcher, and the increasing interest in 3D-printed prosthetics will require standards to ensure that devices meet the expectations of health providers and families, for both upper and lower limb devices. The development of a standard through a body such as ASTM, which coordinates with research institutes to make their database available to researchers, or with a group such as e-NABLE, who will be discussed later, would help reduce costs of independent prosthetic development and promote standardization between designs.
19


Part 2: 3D printing and the Current State of 3D-Printed Prosthetics
3D Printing Hardware. 3D printing, or additive manufacturing, is any method by which objects are created by fusing or depositing layers of material sequentially until the piece is completed (Ventola, 2014). Ventolas paper on the current state of 3D printing applications in medicine has a thorough overview of 3D printing technology. They recognized the three main categories of printers that are commercially available, though many methods of additive manufacturing have been developed since the technologies inception in the 1980s.
The first method of 3D printing discussed by Ventola was Selective Laser Sintering (SLS). SLS printing utilizes a laser passed over a powder-bed of the desired material. This laser heats the powder bed in a precise spot until the particles become fused together. This process is done in the shape of the object, layer by layer until completion (Ventola, 2014). SLS is capable of printing metals, ceramics, and plastics. The resolution of the piece is controlled by the precision of the laser and the size of the powder. Figure 9 illustrates the inner working of an SLS printer, and demonstrates the sintering occurring at the particulate level.
20


Laser scanting erect on
Snlered powder particles (brwn State)
Laser beam
Hrc placed
powder Bed (S*een state)
/*,< V/* - r/ r^v / -'
Unentered material n previous layers
Figure 9 Cross section of SLS print bed and sintering process Wikimedia licensed
under CC BY-SA
The EOS M270 is an SLS machine available during this study through the Childrens Hospital Colorado Research Institute. It has the capability to print in aluminum, Cobalt-Chromium, titanium, and several steel alloys, with a print volume of 228mm x 228mm x 190 mm (EOS, 2017).
The next method of 3D printing Ventola discussed is Fused Deposition Modeling (FDM). FDM printers are the most common and least expensive 3D printers available. They operate by extruding a continuous strand of heated plastic filament out of a print-head, similar to that of a traditional ink printer, onto a print-bed. Like SLS, this print-head moves in the shape of the desired object, layer by layer, until completed (Ventola, 2014). FDM printers allow for greater versatility in their selection of polymers available, and modularity in designs, such as multiple print heads. In Figure 10, a typical FDM 3D printer is illustrated.
21


Figure 10 Schematic representation of the 3D printing technique biown as Fused Filament Fabrication; a filament a) ofplastic material is feeded through a heated moving head b) that melts and extrudes it depositing it, layer after layer, in the desired shape c). A moving platform e) lowers after each layer is deposited. For this kind of technology additional vertical support structures d) are needed to sustain overhanging parts. By Paolo Cignoni, licensed under CC BY-SA
The FDM used during this study was a FlashForge Creator Pro, made available through the University of Colorado Department of Bioengineering.
The last method of 3D-printing highlighted by Ventola is Thermal Inkjet Printing (TB). TB is not a method of 3D printing used for structural components, but rather for tissue reconstruction and bioprinting. Ventola explained the process as follows:
Inkjet printing is a noncontact technique that uses thermal, electromagnetic, or piezoelectric technology
22


to deposit tiny droplets of ink (actual ink or other materials) onto a substrate according to digital instructions.
In inkjet printing, droplet deposition is usually done by using heat or mechanical compression to eject the ink drops. In TU printers, heating the printhead creates small air bubbles that collapse, creating pressure pulses that eject ink drops from nozzles in volumes as small as 10 to 150 picoliters. Droplet size can be varied by adjusting the applied temperature gradient, pulse frequency, and ink viscosity (p. 705).
The limited material volume of TD printing makes it unsuitable for large 3D printed structures, and thus TD was not utilized in this study.
There an additional method of printing utilized in this study not discussed by Ventola, and that is the proprietary photopolymer printing method utilized by Objet Connex printers. These printers deposit a continuous sheet of photopolymer resin onto the print-bed in the defined shape of the object, layer by layer (United States Patent No. 6,259,962 Bl, 1999). This photopolymer is cured by a UV light after the completion of each layer to provide a rigid surface for the next layer to be deposited on. The properties of the photopolymer resins vary based on proprietary compositions, but characteristics of the available resins range from hard plastics, ductile plastics, and rubber-like elastomers (United States Patent No. WO 2004096514 A3, 2004). This method of printing was available at the time of Ventolas publication, so it is unclear why it was omitted.
23


The versatility of materials Objet printers offers makes their use in 3D-printed prosthetics appealing, as parts made multiple material types can be printed alongside each other, while EOS printers are limited to one material type. The material versatility of FDM printers are limited by the number of print heads, typically only one or two. The limited number of printers described by Ventola demonstrates that the literature available to researchers and professionals in the medical field may not cover always cover all available technology. Further work must be done to analyze all available methods of 3D printing in the medical world, not only in the field of prosthetics, but all areas of medicine.
FDM Plastics. The types of materials use in 3D-printers varies widely based on the printer being used, especially in FDM printers, where unlike EOS SLS printer or Objet Photopolymer printers, the user is not restricted by proprietary materials that must be sources from the manufacturer. The two most common plastics used in filament for FDM printers are ABS and PLA. ABS, or Acrylonitrile Butadiene Styrene is a common thermoplastic derived from petroleum byproducts. ABS is considered both strong and ductile, and is used frequently in manufacturing for packaging, toys, and any number of injected molded plastic products (Beginner's Guide to 3D Printing, 2017). As an amorphous solid, it has a glass transition temperature of approximately 105C but no true melting point, making it well suited for 3D-printing (Acrylonitrile Butadiene Styrene (ABS) Typical Properties Generic ABS, 2017). PLA, or Polylactic Acid, is a thermoplastic derived from plant byproducts (Beginner's Guide to 3D Printing, 2017). Like ABS, it is strong but also brittle, making it less suited for use in devices that require durability or resistance to fracture. Unlike ABS, PLA has an organized crystal structure,
24


giving it a glass transition temperature of approximately 57C and a melting point of approximately 161C (Polylactic Acid (PLA) Typical Properties, 2017). Like ABS, this is an ideal temperature range for 3D printing.
More exotic materials have also become commercially available for use in FDM printers. The MarkedForge Mark Two is capable of printing continuous strands of carbon fiber, nylon, Kevlar, and fiberglass impregnated within a polymer filament typical of traditional FDM printers (MarkedForge Materials, 2017). They are advertised as having mechanical properties above that of usual 3D printed thermoplastics, and even above 6061-T6 Aluminum. This claim makes them very appealing for use in prosthetics, where both high strength and light weight are necessary requirements for successful devices.
3D Printed Prosthetics. The lowering costs and availability of 3D printers has made them an appealing choice in manufacturing prosthetics. The widespread use of CAD modeling software among both professional and amateur designers has also led to a wide array of prosthetic designs available online. A widely recognized source of 3D-printed prosthetic designs is e-NABLE. Their organizations mission statement describes themselves as:
[A community of] teachers, students, engineers, scientists, medical professionals, tinkerers, designers, parents, children, scout troops, artists, philanthropists, dreamers, coders, makers and every day people who just want to make a difference and help to Give The World A Helping Hand. (e-NABLE, 2017)
25


e-NABLE acts as a hub where designers can share their files for prosthetic hands that are purpose-made to be created using 3D-printers. The website also offers families of childhood amputees the ability to contact and connect with owners of 3D-printers who wish to produce prosthetic devices for those looking for alternative solutions for their childs disability. At the time of writing, e-NABLE provides files and resources for 7 wrist-actuated designs, also known as tenodesis action, and two elbow actuated designs. The National Institute of Healths 3D Print Exchange also hosts these same designs, and the NIHs database of 3D printed prosthetics is curated by e-NABLE volunteers (3D-Printable Prosthetic Devices, 2017).
Very little clinical evaluation of the 3D-printed devices available from e-NABLE has been conducted. Zuniga et al. examined one design available through e-NABLE, the Cyborg Beast. In their study, Zuniga et al. proposed a method for sizing and fitting the hand appropriately for a user remotely, so that the hand may be delivered to the user ready-to-use, without further adjustment needed by the family. They found that their method of remote fitting was successful, but that there is a need for further studies that examine functionality, validity, durability, benefits, and rejection rates of 3D-printed hands (Zuniga, et al., 2015).
A review of 3D-printed upper limb prosthetics by Kate et al. included 5 e-NABLE designs out of a total of 58 devices. While thorough, this review only examines the mechanical and kinematic specifications provided by the designers. The study goes into great detail on aspects such as the range of motion of individual joints, number of joints, degrees of freedom, extension/flexion methods, type of control etc. This data may not be useful to the end user, as it is not presented in a manner that helps a clinician or
26


occupational therapist in selecting a device, and none of it is verified through actual testing and physical evaluation of the hands. While Zuniga et al. provided a sound method for fitting, and Kate et al. provided detailed technical data on designs, none of it provided information on how the devices performed, how patients should use them, what sort of durability is expected, or if patients will even accept them. This is the gap in knowledge on 3D-printed hands this study will address.
SPECIFIC AIMS
From the literature review, it is clear that 3D printing can help solve many issues with current prosthetic technology, mainly cost, accessibility of devices, and functionality with aesthetic appeal. However, to date there is little work done on how 3D-printing can best be applied to pediatric prosthetics. There is no uniformity in information on which printers and materials are best used in prosthetics. There is no easily accessible information for specialists for which of the dozens of open-source 3D printable hands available provide the best features or designs that would best suit their patients. Finally, there is no easily accessible guide for specialists on how the available open-source designs should be constructed, starting from the printing stage, to post-possessing, and final assembly. This study will address these gaps by providing the preliminary steps needed to create standards which can be utilized by specialist in the field of pediatric prosthetics to best serve their patients and answering exactly how 3D-printed prosthetics can best be used. The following specific aims will be addressed:
Specific Aim 1: Document and evaluate printing and construction considerations of currently available open-source 3D printed upper-limb prosthetics.
27


Specific Aim 2: Determine the Suitability of Structural Plastics used currently in 3D printing for use in upper-limb prosthetics.
Specific Aim 3: Evaluate hand designs for usability by the patient.
II. EXPERIMENT AND RESULTS
Experimental Set-up
Construction of 3D-printed hands. With regards to Specific Aim 1, the CAD files for the hands evaluated in this study were all obtained through e-NABLE. The Raptor Hand, Raptor Reloaded, Cyborg Beast, and Flexy hand were chosen for evaluation at various steps of the study. Renders of these hands are shown in Figure 7.
28


Figure 7- a) Raptor Hand Render by Jeremy Simon is licensed under CC-BY 3.0 b) Raptor Reloaded c) Flexy Hand d) Cyborg Beast Render by Creighton Labs is licensed
under CC BY-NC
Inclusion and Exclusion Criteria for 3D-Printed Hand Designs. The Raptor and Raptor Reloaded were chosen to examine two iterations of the same basic design for improvement. The Flexy Hand was chosen due to its use of live-hinges, rather than pinned hinges, and its natural prehensile thumb design. The Cyborg Beast was selected due to its use occurrence in the literature as well as its use of non-3D-printed parts. For all testing, all components of the Flexy Hand were produced at 70% scaling to represent
29


the dimensions of a device intended for a child. All other hands were left at their default
scaling, as they were suitably sized for use by children at 100% scaling. When applicable, all components for the hands were printed for a right-hand user. For consistency in testing, no left-hand designs or variants were utilized during the study. Designs available outside of e-NABLE were excluded in order to keep the scope of the experiment concise and ensure the hands evaluated will be accessible for future studies. All files are available in .stl format, which is accessible by the majority of CAD software packages, including Solidworks, AutoCAD, and Microsoft 3D Builder.
All hands selected required additional manufactured hardware to be built, none of the selected designs could be built solely from 3D printed materials. All hands required wood screws for the anchoring and adjustment of tension pins that secure the tension lines and kept the flexion lines taught. Braided high strength fishing line was used in all hands to act as tension lines for the generation of finger flexion motion. Braided elastic string was used for finger extension when the force generated by the wearer in wrist flexion was ceased. The Flexy Hand did not require the use of elastic string for finger flexion, as the polyurethane live-hinges return the hand to an open position when force generated by the wearers wrist is ceased. The Cyborg Beast requires the addition of a metal Chicago bolt to pin the proximal finger pieces to the palm, rather than a printed pin (Raptor, Raptor Reloaded) or interference socket (Flexy Hand) that was used in the other hand designs.
A fifth design, the Phoenix Hand v2, was initially included in the evaluation process (Figure 8).
30


Figure 8 Phoenix Hand v2 by EnableCommunityFoundation is licensed under CC BY-
NC
The Phoenix Hand was chosen due to its similar design and construction to the Raptor hands, with mechanical hinges and pins, but features an adducted and opposed thumb similar to the Flexy Hand. The design also featured enclosed tracks that the flexion lines needed to be fed through to reach their tension pin anchors. However, the addition of supports during the 3D-printing process filled these tracks and subsequently could not be removed from such a small enclosed space. Attempts to reorient the part on the build-platform to alleviate the need for support in the flexion line tracks resulted in
31


failure, as a piece as large as the palm could not be supported by the two small, round wrist hinges (Figure 9).
Figure 9 Failed Pheonix Hand v2 print due to insufficient surface area for proper
filament adhesion.
The palm of the Flexy Hand uses the same method of guiding its flexion wires, but the rear of the palm is flat, allowing it to be printed upright easily, thus solving the issue of removing support from within the flexion line tracks. Given the large amount of material that would be needed in the trial and error process needed to solve this issue, the Phoenix Hand was eliminated from the study.
Printers and Software. For the completion of Specific Aim 1 and 2, two FDM printers were utilized to construct all components for mechanical testing purposes were FlashForge Creater Pros by FlashForge USA, City of Industry C A. This printer is a dual print-head design with a heated build-platform, and enclosed build-chamber for ambient temperature stability. One printer was modified with a Dual Flexion flexible filament extruder by Diabase Engineering, Longmont CO. This allowed for printing of
32


NinjaFlex polyurethane filament to be used in construction of the live-hinge components of the Flexy Hand.
The control software for the FDM printers used at the beginning of the study was MakerBot Desktop. This package is available free of charge from 3D-printer manufacturer Makerbot, Brooklyn NY. The software allows for the import of .stl files and converts them into G-code, which is read by the printer as a series of sequential x-y-z coordinates to guide the print-heads into the proper shape of the desired object. The software package also allows for control of print-head and print-bed temperatures depending on the selected plastic, as well as options to adjust infill percent, part scaling, place support material, rafts, and adjust the quality of the print. Test prints with MakerBot Desktop resulted in components that were significantly undersized from their expected dimensions. The software also could not produce rafts of consistent quality that allowed the plastic to adhere properly to the print-bed. Attempts to address these issues by changing print parameters within the software were unsuccessful. Control software for the FDM printers was then changed to Simplify3D, by Simplify3D, Inc., Blue Ash OH. Simplify3D is a license-based 3D printer control software package with more features than MakerBot Desktop. The switch to Simplify3D corrected all issues in test prints that came from the use of MakerBot Desktop.
All photopolymer printing was done on an Objet Connex350. This printer includes its own proprietary control software that adjusts temperatures and UV exposures automatically depending on the selected materials. CAD files are imported and the user can control scaling, placement on the print bed, and build quality.
33


Selection of Plastics. With regards to Specific Aim 2, ABS was selected for examination on all hand parts and material test specimens. PLA was excluded due to its brittleness and tendency to fracture under stress rather than deform. Prosthetics in practical use are dynamic devices and there is the expectation from the wearer that the device will be robust and able to stand-up to everyday wear and tear, as well as minor bumps and drops. PLA does not allow for this expectation to be met, thus its exclusion.
Two brands of ABS were used to examine if variation in filament manufacturer would need to be a consideration of clinicians producing 3D-printe devices. A generic brand of filament previously available in the laboratory space was used, as was a brand-name ABS filament from Hatchbox. Hatchbox was selected due to its competitive price, and ease of availability from multiple retailers. These are parameters that would be a significant consideration by clinicians when selecting which filaments to purchase for the construction of 3D-printed devices. Generic ABS and Hatchbox ABS were used to construct tensile specimens. Only Hatchbox ABS was used to construct full hands and parts for usability testing, due to the limited quantity of Generic ABS available in the lab-space. NinjaFlex by NinjaTek was used to construct the elastomer live-hinges for the Flexy Hand. NinjaFlexs mechanical properties were not examined in this study.
All Generic ABS samples were printed with the extruder at 230C and 100C build platform. All Hatchbox samples and device parts were printed with the extruder at 226C and 106C build platform. The extruder speed for both ABS examples was 3600 mm/min. All prints done in ABS were done with the print area enclosed, with the printer door closed and the lid on, as to provide a stable ambient temperature. All NinjaFlex parts were printed with a 230C extruder and a 40C print bed, and with a print speed of
34


1500mm/min, per the manufactures recommendations (NinjaTek, 2017). Prints in NinjaFlex were done with the printer door closed but with the lid removed, to promote faster cooling. All prints utilized the default raft generated in Simplify3D. Supports on all prints were added using the automated support generator feature in Simpify3D.
All parts made on the Objet Connex350 were built with RGD525. Objet advertises this photopolymer as having ABS-like properties and was compatible with the available Objet Connex350 printer, thus a comparison between actual ABS filament is warranted (Stratasys, 2017). This photopolymer was used to construct tensile specimens as well as hands for usability testing.
For a basic comparison of available composite 3D-printed materials relative to ABS and RGD525, single tensile specimens were sourced from outside vendors constructed in carbon-fiber-reinforced and Nylon-reinforced plastics from a Marked Forge 3D-printer. The 3D-printed carbon-fiber testing specimen was sourced from I Heart RC Hobby, Salt Lake City, UT, and the 3D-printed Nylon-reinforced specimen was sourced from ProductGoGo, Littleton, CO (I heart RC Hobby, 2017; ProductGoGo,
2017).
Tensile Testing. To complete Specific Aim 2, tensile testing on both brands of ABS, RGD525, and the carbon-fiber and Nylon reinforced samples was done in accordance with ASTM D638 (ASTM International, 2014). All tensile specimens were printed to meet the specified Type 1 dimensions for rigid plastics. The dimensions of the ASTM Type 1 tensile specimen are shown in Figure 10.
35


Figure 10 ASTMD638 Type 1 Tensile Specimen W=13mm; L=57mm; W0=19mm;
L0=165mm; G=50mm; D=115mm; R=76mm; T=3.2mm. Wc and WO are defined in ASTMD638for alternate specimen types or specific materials not utilized in this study
(ASTMInternational, 2014).
Tensile specimens made in ABS were printed at 100% infill to achieve the highest possible strength. Tensile specimens printed in RGD525 were printed with the Objets default infill setting. The single carbon-fiber specimen was received with the printed fiber strands oriented parallel to the specimens longitudinal axis. The single Nylon specimen was received with Nylon fibers oriented along the perimeter of the specimen. The as-received carbon-fiber and nylon tensile specimens are shown in Figure 11.
Figure 11 Carbon-Fiber (left) and Reinforced Nylon (right) Type 1 Tensile Specimens.
36


Two versions of ABS and RGD 25 tensile specimens were produced to examine the effect a parts orientation on the print bed ha on the resulting mechanical properties. These two orientations were designated as Flat and Vertical, and represented a 90 rotation along the tensile specimens longitudinal axis. The Flat print orientation is shown in Figure 12a, and the Vertical Print orientation is shown in Figure 12b.
Figure 12 a) Flat print orientation b) Vertical print orientation.
The first set of vertical and flat RGD525 specimens and all Generic ABS tensile
samples were tested on an MTS Insight tensile frame. All remaining tensile specimens were tested on a Mark-10 ESM1500 tensile frame. This change was made to increase the volume of tests that could be performed, as the Mark-10 ESM1500 was more readily accessible to the researcher. Readings from load cells on both machines were measured and recorded in newtons, and displacement was measured and recoded in millimeters.
All tensile specimens were pulled at a speed of 5mm/min. Results of tensile tests were
37


rejected if the specimen fractured outside of the designated gauge length, per ASTM D638, with the exception of the Carbon-fiber and Nylon reinforced samples, as only one was available.
Lateral Testing for Standard Development. To standardize mechanical testing for 3D-printed prosthetics and give a measure of usability outlined in Specific Aim 3, a universal method of testing hands for relative durability in an easily reproducible and common mode of failure was needed. A lateral force on the fifth digit was seen to be a typical location for high and rapid loading, such as if the wearer were to stand up from a seated position and support their body weight on the hand. This motion was replicated using a Mark-10 ESM1500 tensile frame with custom designed mounting brackets for mounting and imparting force on test-sections of each hand design. The hands tested in this mode of failure were the Raptor, Raptor Reloaded, Cyborg Beast, and Flexy Hand. The palm of each hand was modified into a test jig using Microsoft View 3D to reduce the part down to a quartered section containing the mounting position for the fifth digit. Renderings of these test jigs are shown in Figure 13.
38


Figure 13 a) Raptor test jig b) Raptor Reloaded test jib c) Cyborg Beast test jig d) Flexy
Hand test jig.
Attempts to modify the CAD files for the selected designs were originally made in SolidWorks 2015, but the high polygon count of these parts was not easily convertible by the software from .stl to the proprietary .sldprt file format. The Flexy Hand palm .stl file for example, has a surface comprised of over 80,000 polygons. This caused software crashes and hardware freezes regularly, and no progress could be made. The switch to Microsoft View 3D solved this issue, as the software does not require the conversion of the .stl to a proprietary file format for edits to be made, and the part is rendered as a wireframe model rather than a solid body, reducing the computing power needed to render the models.
39


Mounting brackets for holding the modified palms and fifth finger were assembled in the proper position to simulate a lateral load and a hook to pull the digits that could be held by the tensile frame grips were designed in SolidWorks 2015. Renders of the mounting bracket and tensile hook are show in Figure 14.
Figure 14 a) Test jig mounting bracket b) Lateral Failure Mode tensile hook.
Both the mounting bracket and the tensile hook were printed on an EOS M270
metal 3D-printer in maraging steel for strength properties that would far exceed the plastics being tested, as to avoid failures from the testing equipment. Support removal and surface finishing of the steel components included a combination of hand held electric grinders, end milling, and sand blasting.
Both testing components have a large, thin area for inserting into the tensile test grips of the Mark-10 ESM1500. This provided sufficient surface area for the grips to latch on to so the testing components did not come loose during the lateral testing procedure. Mounting holes were drilled in the modified palms and mounting bracket so they could be quickly swapped in and out and were secured by bolts. The hook for
40


pulling the fifth digit was placed at the joint between the proximal phalange and intermediate phalange for each trial. The complete testing setup is shown in Figure 15.
Figure 15 Later ail Failure Mode Test Setup (post break).
Readings from load cells on both machines were measured and recorded in
newtons, and displacement was measured and recoded in millimeters. The speed at which lateral testing occurred was done at 5mm/min, lOOmm/min, 500mm/min, and lOOOmm/min. Speeds were chosen to represent the lowest loading speed, as designated by ASTM D638, and increased upward through 2 orders of magnitude. Varying the loading rate was done to determine if parts constructed through 3D printing exhibit a strain-rate dependence, a common mechanical property of plastics. All parts were printed
41


at 40% infill to represent typical 3D-printer parameters, which are usually selected to reduce the overall amount of material used, and thus reduce cost.
Usability Modified SHAP test. For the completion of Specific Aim 3, a modified SHAP test was performed on the Raptor, Raptor Reloaded, and Flexy Hand to determine what tasks and what prehensile patterns were easily completed by the 3D-printed devices. SHAP tasks were not timed, as the purpose of the examination was to determine which tasks were achievable, not how quickly they could be achieved. As shown by Vasluian et al, the SHAP test is inherently flawed when performed by children, since all normative values used in its score calculations are intended for adults. Thus, all SHAP scores would be invalid. Hands were manipulated by a non-limb deficient user in the seated position defined by SHAP parameters. Each task was evaluated under a pass/fail criteria if the task could be completed under the parameters of the SHAP test. Care was taken not to orient the device in such a way that would be considered unnatural, and only in a manner in which a typical user could achieve i.e. beyond 90 perpendicular to the users seated position and the test area, or inverted. All SHAP tasks were tested except Lifting Heavy Object, as the volume of water required to fill the carton would have proved a hazard to the electrical equipment in the researchers shared lab-space. Initial attempts at performing tasks revealed that no objects could be manipulated, as the smooth plastic of the fingers did not create enough friction for objects to be securely held. All hands were fitted with Lee Tippi Micro Gel Fingertip Grips for final testing so that objects would not slip as easily while being manipulated (Figure 16).
42


Figure 16 Raptor Hand with Lee Tippi Micro Gel fingertips.
Grip strength tests was attempted during the study, using a Camry EH101 hand
dynamometer and the 3 hands selected for use in the SHAP task evaluations. The dynamometer is shown in Figure 17.
Figure 17 Canny EH 101 hand dynamometer.
43


It was discovered that the dynamometer could not be fit into the hand in any position that allowed for consistent and repeated grip strength to be measured. Flexing the hands against the dynamometer handle regularly caused the flexion line knots to untie and loosen, and the tension pins to pull out from the anchoring screws. Due to these repeated failures, data could not be consistently collected. This test was not pursued further in this study.
Results
Tensile Testing. Engineering stress and strain of the Generic ABS in both flat and vertical print orientations are shown in Figure 18. Each line represents one specimen trial (n=3).
Generic ABS Flat Generic ABS Vertical
Strain Strain
Figure 18 Stress Strain curves of the Flat and Vertical Generic ABS tensile specimens.
44


Engineering stress and strain of the Hatchbox ABS in both print orientations are shown in
Figure 19 (n=6).
Hatchbox ABS Flat Hatchbox ABS Vertical
Strain Strain
Figure 19 Stress Strain curves of Flat and Vertical Hatchbox ABS tensile specimens Engineering stress and strain of RGD525 in both print orientations are show in Figure 20
(n=7).
525 Flat 525 Vertical
Strain Strain
Figure 20 Stress Strain curves of Flat and Vertical RGD525 tensile specimens.
45


The mean ultimate tensile strengths(UTS) from all trials, from all three materials, in both print orientations is show in Figure 21.
Figure 21 Mean Ultimate Tensile Strengths from all specimens and print orientations
with standard error.
46


A comparison of the mean UTS of the tested plastics with the expected UTS values from the literature as well as the results of the single-sample 3D-printed composites are shown in Figure 22.
Mean Ultimate Tensile Strengths of Selected Materials and Composites
Figure 22 Mean Ultimate Tensile Strengths from all specimens and print orientations with expected literature values and composite samples.
Statistical analysis of the tensile test results was performed using 2-group Student t-tests between each set of test results: first, comparing the effect of materials within orientation groups, and then comparing the effect of orientation within material groups. The resulting p-values are shown in Table 1.
47


Table 1 Student t-test p-values of tensile tested plastics. Upper left Comparing plastic types in Flat orientation. Lower left Comparing plastic types in Vertical print orientation. Right Comparing Flat us Vertical Print orientations for each tested plastic.
p-value p-value
FLAT FLAT v. VERTICAL
Generic v. Hatch 0.2596 Generic ABS 0.0116
RGD525 v. Generic 4.81E-05 Hatchbox ABS 0.0123
RGD525 v. Hatch 5.58E-06 Photo. 525 0.471
VERTICAL
Generic v. Hatch 3.32E-04
RGD525 v. Generic 2.60E-03
RGD525 v. Hatch 3.77E-04
No statistical significance in resulting UTS is shown between the Generic ABS and Hatchbox ABS in the flat orientation (p>0.05). All other material comparisons when controlling for print orientation on resulting UTS show statistical significance (p<0.05). Print orientation is shown to be statistically significant in resulting UTS for booth Generic ABS and Hatchbox ABS (p<0.05). Print orientation is shown not to be statistically significant in resulting UTS for RGD525 (p>0.05).
Lateral Failure Mode Testing. Force-deflection plots of the Cyborg Beast hand in the purposed Lateral Failure mode test at 5mm/min, lOOmm/min, 500mm/min, and lOOOmm/min strain rates are shown in Figure 23. Results of the same test on the Flexy Hand, Raptor, and Raptor Reloaded, are shown in Figure 24, 25, and 26, respectively.
48


Force Kg Force Kg
Cyborg Beast 5 mm/mln Cyborg Beast 100 mm/mln Cyborg Beast 500 mm/mln Cyborg Beast 1000 mm/mln
- 70 60 70 60 70 60 -
- 50 50 50 -
- :orce Kg C. o :orce Kg 4k o =orce Kg 4k o -
30 30 30
- 20 20 20 -
. 10 10 10 _
£ i i 0 0 0
10 20 30 0 10 20 30 0 10 20 30 0 10 20
Deflection mm Deflection mm Deflection mm Deflection mm
Figure 23 Cyborg Beast Lateral Failure Mode Test results
Flexy Hand 5 mm/min
Flexy Hand 100 mm/min
Deflection mm
Flexy Hand 500 mm/min
Flexy Hand 1000 mm/min
Deflection mm
Figure 24 Flexy Hand Lateral Failure Mode Test results
49


Force Kg Force Kg
Raptor 5 mm/min
Deflection mm
Raptor 100 mm/min
Raptor 500 mm/min
Deflection mm
Raptor 1000 mm/min
Figure 25 Raptor Lateral Failure Mode Test results
Raptor Reloaded 5 mm/min Raptor Reloaded 100 mm/min Raptor Reloaded 500 mm/min Raptor Reloaded 1000 mm/min
Figure 26 Raptor Reloaded Lateral Failure Mode Test results
50


Fracture Strength (Kgf)
The mean failure strength of each hand at the four strain rates is shown in Figure 27.
Mean Fracture Strength in Lateral Pull
50
45
40
35
30
5 100 500 1000
Strain Rate (mm/min)
FlexyHand Cyborg Beast Raptor Reloaded Raptor
Figure 27 -Mean Failure Strengths of the tested hand designs versus strain rate with
standard deviation.
Statistical analysis of the lateral failure mode test began with a one-way ANOVA
of the mean failure strength of each hand design at each strain rate. The ANOVA Tables, statistics, and Tukey HSD results for each hand are shown in Figures 28-39. A post-hoc Tukey Honest Significant Difference test was then performed on the ANOVA results
51


from each hand design to determine if the mean failure strengths between each strain rate were significantly different.
Figure 28 Raptor ANOVA Table ANOVA Table
Source SS df MS F Prob>F A
Columns 175.38 3 58.461 0.39 0.7584
Error 2368.86 16 148.054
Total 2544.24 19
Figure 29 Raptor ANOVA table statistics
52


Raptor Tukey HSD



u

5 10 15 20 25 30 35 40
No groups have means significantly different from Group 1
Figure 30 Raptor Tukey HSD results
Raptor Reloaded ANOVA Table
Figure 31 Raptor Reloaded ANOVA Table
53


ANOVA Table
Source SS df MS F Prob>F
Columns 90.59 3 30.1966 1.32 0.3022
Error 365.687 16 22.8554
Total 456.277 19
Figure 32 Raptor Reloaded ANOVA Table statistics
1
2
3
4
6 8 10 12 14 16 18 20 22
Raptor Reloaded Tukey HSD
-------1------1-----1-------
---Q----------------
No groups have means significantly different from Group 1
Figure 33 Raptor Reloaded Tukey HSD results
54


Cyborg Beast ANOVA Table
Figure 34 Cyborg Beast ANOVA table ANOVA Table
Source SS df MS F Prob>F a
Columns 0.2773 3 0.09243 0.06 0.9803
Error 12.6998 8 1.58747
Total 12.9771 11 V
Figure 35 Cyborg Beast ANOVA table statistics
55


Cyborg Beast Tukey HSD
3 3.5 4 4.5 5 5.5 6 6.5 7 7.5
No groups have means significantly different from Group 1
Figure 36 Cyborg Beast Tukey HSD results Flexy Hand ANOVA Table
Figure 37 Flexy Hand ANOVA table
56


ANOVA Table
Columns 0.12131 3 0.04044 3.89 0.0553
Error 0.08319 8 0.0104
Total 0.2045 11 V
Figure 38 Flexy Hand ANOVA table statistics
1
2
3
4
1.4 1.5 1.6 1.7 1.8 1.9 2
Flexy Hand Tukey HSD
-e-------------
-e-
The means of groups 1 and 3 are significantly different
Figure 39 Flexy Hand Tukey HSD results P-values from the resulting ANOVA on each hand design demonstrated there was
not a statistically significant difference between the resulting mean failure strength and strain rate (p>0.05). The post-hoc Tukey HSD determined that there was no mean failure strength significantly different regardless of strain rate in all cases except one: the Flexy Hand, displayed a statistically significant difference in means between the failure strength at strain rates of 5mm/min and 500mm/min.
SHAP Task Results. The pass/fail results of each hand evaluated in the
completion of the 25 SHAP tasks is shown in Table 2.
57


Table 2 SHAP Task results from each tested hand on pass fail basis and primary prehensile pattern utilized for each task.
Prehensile Flexy Raptor
Task Pattern Hand Raptor Reloaded
Wood Extension Extension pass pass pass
Metal Extension pass pass fail
Page Turning fail fail fail
Carton Pouring fail pass pass
Lifting Light
Object fail pass pass
Wood Lateral Lateral pass fail fail
Metal Lateral pass fail fail
Button Board fail fail fail
Lifting Tray pass pass pass
Rotate Key fail fail fail
Wood Cylinder Power pass pass pass
Metal Cylinder pass pass pass
Glass Jug
Pouring fail pass fail
Door Handle pass pass pass
Wood Sphere Spherical pass pass pass
Metal Sphere fail fail fail
Jar Lid fail fail fail
Wood Tip Tip pass pass fail
Metal Tip pass pass fail
Pick up Coins fail pass pass
Open/Close Zip fail fail fail
Wood Tripod Tripod pass pass pass
Metal Tripod pass pass pass
Food Cutting fail fail fail
Rotate Screw fail fail fail
Results were further broken down into how many tasks, based on the primary prehensile pattern, each hand completed. A percentage value of the total number tasks the hand was able to complete was calculated to determine an overall Task Completion Rate. These results are shown in Table 3.
58


Table 3 Overall Completion rates from each hand design based on primary prehensile
pattern
Task Completion
Spherical Tripod Power Lateral Tip Extension Totals Rate
Flexy 1 2 3 3 2 2 13 52%
Raptor 1 2 4 1 3 4 15 60%
Raptor Reloaded 1 2 3 1 1 3 11 44%
During the course of the SHAP task evaluations, all hands suffered from mechanical failures. The Flexy Hand had the 5th digit tension pin pull out from the anchoring screw, and the 4th digits flexion line knot came undone. The Raptor and Raptor Reloaded both experienced these same failures during the test. The test was stopped for repairs to be made on the hands, and resumed once the hands and been fixed.
Discussion
Construction Observations. The use of 3D-printers for quickly manufactured, and inexpensive prosthetics is possible, but there are many considerations that were recognized in this study that are not found in existing literature. The set-up and maintenance of common FDM 3D-printers themselves is a non-trivial requirement for any clinic that would want to produce 3D-printed devices for their patients. In the course of this experiment, build-plate platters needed to be replaced, and print-head assemblies were upgraded to meet new material requirements. Filament mis-feeding was also a regular occurrence during filament swaps, and each material replacement called for a trial and error phase where both print-head and build-plate temperatures needed to be discovered for optimal part adhesion and build quality with each material examined. All these parameters are printer-dependent, and no two 3D printers behave identically.
59


Diagnosing these problems required familiarity not only with 3D-printing technology, but with each unique printer. A plug-and-play approach to 3D-printers in a clinical setting would result in poor part quality and printer down-time as problems persist. Trained staff to maintain and operate 3D-printers would be a requirement for 3D-printed prosthetics to become commonplace.
The software that controls the parameters and feeds instructions to the printer itself is complex and requires experience to be able to use it to its fullest extent and produce the highest quality print. Parameters commonly modified during this study included support type, infil type, infil density, extruder and build-platform temperature, inclusion or exclusion of rafts, and print-head speed. Fine-tuning these parameters must be done by trial-and-error, and are dependent on the material being used. Changing materials requires these parameters be changed as well. For example, Hatchbox ABS is advertised to require an extruder temperature in a range of 210C-240C, and a build-platform temperature range of 55C-85C. Likewise, NinjaFlex designates an extruder temperature range of 225C-235C, and a build-platform temperature of 40C. Trial and error during this studied showed that an extruder temperature of 226C and build-plate temperature of 106C was ideal for Hatchbox ABS, and 230C extruder temperature and 40C build-platform temperature for NinjaFlex. Simplify3D allows for material profiles to be created by the user and quickly interchanged, but these profiles must be made through a trial-and-error process.
Complications with the Makerbot Desktop software that led to the switch to Simplify3D showed that available software packages were not innately reliable and their accuracy must be evaluated by the user. The unintended part-scaling that was discovered
60


during this study may not have been found by a less-experienced user until parts had already been delivered, such as to a clinician and patient for final fitting. At that point, the problem would have to be diagnosed and the process restarted, delaying the patient from receiving the care they expect from the clinician.
Zenga et al.s study on a fitting process for the Cyborg Beast states inaccurate scaling or significant errors in the measurements could affect the function or fitting of the 3D-printed prosthetic hand., but it does not explain what these errors or difficulty in function may be. If attention by the user is not taken, it is possible for a part to be scaled below the resolution of the printer in use, resulting in a zero-dimension feature that will fail to print properly. Alternatively, a part may be scaled down so small that robust features at 100% scaling may no longer have structural integrity. The issue of scaling is one that may not be readily apparent to prescribers of 3D-printed devices.
Errors in the files provided through e-NABLEs online database were also discovered during this study. The individual phalanges for the construction of the Flexy Hand can be built using an included finger plate which has all the individual pieces pre-assembled into one file so that the user does not need to go through the task of counting and scaling each individual phalange for all five fingers. A CAD rendering of this part is shown in Figure 40.
61


Figure 40 Render of Flexy Handfinger build plate.
However, this file contained an error in which the pieces were not zeroed to be
flat on the build-platform. The thin geometry of the part in the orientation designated in the file resulted in little surface area for adhesion to the underlying support to fill the space between the part and the build-platform. This resulted in numerous attempts where the printer ended up printing in air, where the underlying part had failed to adhere properly and the extruder continued to travel upward. The resulting failed print of the Flexy Hand finger-plate is shown in Figure 41.
62


Figure 41 Print failure of Flexy Hand printer build plate due to improperly constructed
CAD file.
Only after this error was discovered and corrected could the Flexy Hand be properly constructed. An inexperienced user of the software and in printing these devices maybe not recognize this error until after multiple failed prints have occurred, and maybe not have access to proper CAD software to edit and correct the error in the e-NABLE file.
Under-designed features are prevalent in all hand-designs examined. Common among all designs were the use of tension pins for supporting the braided fishing that acts as flexors tendons. The pins from each design have walls too thin to properly serve as thread anchors for the designated screws. The Raptor hand had a tensioner pin wall thickness of only 0.18mm, the Raptor Reloadeds tension pin wall thickness was only
63


0.70mm, and the Flexy Hands was 0.87mm, at the 70% child-sized scaling used in this study. As a result, pins splitting when screws were tightened for proper flexor tension resulted regularly. This is shown in Figure 42.
Figure 42 Example of tension pin splitting common among all designs tested. Another common design feature shared among all the assembled hands was the
wrist gauntlet, which provides cut-outs for strapping and attachment to the wearers arm During the course of the study, the gauntlets for both the Flexy Hand and Raptor Hand fractured through no fault but regular wear and use. These are shown in Figure 43.
Figure 43 Gauntlet failures of Flexy Hand (a) and Raptor (b). Also note the tension pin block split in the Flexy Hand gauntlet from tightening offlexion lines.
64


The thin features of these components need to be enhanced in order to provide a robust and durable device for the patient, otherwise repeated failures may quickly lead to device abandonment.
Tensile Test Results. While the difference in UTS for the flat and vertically printed ABS tensile specimens was statistically significant, the practical difference was small. The mean UTS for the flat Hatchbox ABS and vertical Hatchbox ABS was 29.5 MPa and 34.5 MPa, respectively. This is only a 14.6% difference in mean UTS between Hatchbox ABS print orientations. The mean UTS for the flat Generic ABS and vertical ABS was 26.5 MPa and 30.4 MPa, respectively. This is only a 12.8% difference in the mean UTS between Generic ABS print orientations. There was no significant statistical between Hatchbox ABS or Generic ABS when printed in the flat orientation. While the vertical orientation was statistically significant, the practical difference again was small, with a difference of only 11.8%. These results suggested that contrary to the literature, print orientation was not a significant factor in the strength of ABS parts made for 3D-printed prosthetics.
The effect of print orientation was shown to be significantly insignificant in the specimens printed in RGD525. The mean UTS for the flat RGD525 and vertical RGD525 was 49.4 MPa and 52.1 MPa, respectively. This is only a 5.2% difference in mean UTS. In the application of 3D-printed pediatric prosthetics, which are created with the intent of being inexpensive, and easily replaceable, this small difference in mean UTS between the examined print orientations indicated this does not need to be a major consideration.
65


The standard deviation of the flat and vertical RGD525 specimens was 4.7 MPa and 8.4 MPa, respectively, which was higher than he standard deviations of both Hatchbox ABS(Flat-3.9 MPa, Vertical 1.0 MPa) and the Generic ABS(Flat-1.5 MPa, Vertical 0.4 MPa). This demonstrated the resulting mechanical properties of devices created using Objet photopolymer printers will not have reliable mechanical properties as compared to those printed in ABS from an FDM printer.
The stress-strain curves of the two tested material types showed a distinct difference in modes of failure. While both the ABS samples and RGD525 samples exhibited similar elongations of 12%-22%, their failure modes differ. The distinct region of plastic deformation past the UTS shown in both the ABS sample results indicated higher ductility than RGD525, which show almost no region of plastic deformation, and thus brittle failure (Figures 18, 19, 20). Despite the higher UTS seen in RGD525, its wider range of expected mechanical properties and tendency to fail in a brittle manor makes it less suited for 3D-printed prosthetic devices than ABS. While ABS had a lower UTS, it is more ductile, and thus any loading beyond its UTS will not result in immediate and complete fracture of a part. This robustness makes ABS the desired material.
The comparison of the tested materials to the material properties found in the literature showed that expected results for both the ABS and RGD525 were achieved. Comparing the two tested materials to the single-trial composite samples showed that Nylon-re-enforced samples were not necessarily stronger than ABS or RGD525, with a UTS of only 65.0 MPa. This was only 19.9% higher than the vertical RGD525 samples, the strongest result observed in this test. However, the tested carbon-fiber sample displayed strengths on an order of magnitude higher than the tested plastics, at a UTS of
66


224.1 MPa (Figure 22). The increased cost of the Marked Forge printers capable of producing parts in carbon-fiber goes against the idea that 3D-printed prosthetics are an inexpensive alternative to traditional prosthetics. A MarkedForge Mark Two printer with composite printing capability costs $13,499, and a spool of their proprietary carbon-fiber filament is $447 for a 150cm3 build volume (MarkedForge Materials, 2017). Currently, the overall cost of producing 3D-printed composite prosthetics surpasses that of traditional devices. However, if the technology were to become less expensive and more widely available, they would be the superior choice over either ABS or RGD525.
Lateral Failure Mode Testing. The results of the lateral failure mode test developed for this study are presented as the force applied to the test pieces, in kilograms-force, versus how far the finger deflected before fracture, in millimeters. These values were chosen over the standard stress and strain values of pascals and mm/mm in order for them to be more clearly understood by amputee clinicians and specialists who are less familiar with engineering principles. Kilograms-force was chosen over pascals, as it is a more accessible measure of mass relative to the force of gravity, rather than pressure.
The SI unit of force in similar tests is newtons. However, in keeping with the premise that this test is to be a standard for those unfamiliar with engineering principles, the more commonly used unit of kilograms-force was chosen. In a wider application, this could be further localized into imperial units of pounds-force. Furthermore, a unit of force was maintained rather than pressure, as it broadens the accessibility of the failure mode test to many different device designs without having to know the cross-sectional area of various components of the pieces being tested, which would be required for the accurate calculation of applied pressure. In the case of the hands tested, the complex geometries
67


of the proximal phalanges, the metacarpal joints, and their fixation pins would make this a non-trivial process. Strain, which is often considered unitless, is a confusing measure to those unfamiliar with engineering principles. Deflection in millimeters provides an accessible and relatable measure of how far a component would have to be bent before fracture. Again, this also requires no previous calculation of gauge length, which when testing complex shapes, is non-trivial to derive. In wider applications, the measurement of deflection could be presented in imperial units of inches or fractions of inches.
The Raptor showed the highest consistent failure strengths, 16.2-24.1 kgf (Figure 25), of all the tested designs, with the Raptor Reloaded being the second highest, at 11.3-16.7 kgf (Figure 26). This is due to the Raptors much larger and robust design, whereas the Raptor Reloaded has been slimmed down and dimensions reduced for a more aesthetically pleasing form. The Cyborg Beast strengths ranged from 5.1-5.5 kgf (Figure 23), which performed marginally better than the Flexy Hand. The inclusion of the Cyborg Beast in this study was to examine how hand designs with sourced metal hardware performed compared to fully 3D-printed designs. What was shown was that the addition of metal fixture proved no better at improving mechanical strength than their 3D-printed counterparts. The use of metal hardware goes with the assumption that the weak-point of 3D printed hands are the fixture pins. This study showed that not to be the case. Rather, all tests run in the lateral failure mode had failure occur at the rear pinhole of the proximal phalange, and no failures of the pin itself occurred on either the Raptor or Raptor Reloaded (Figure 44).
68


Figure 44 Proximal Phalange joint failure experienced in all lateral mode failure tests. From left to right: Cyborg Beast, Raptor Reloaded, Raptor.
The pin hole of the Raptor and Raptor Reloaded was 5.2mm wide, the Cyborg Beasts was only 3.0mm wide. The Cyborg Beast experienced much lower failure strengths due to its dimensionally thinner components. The pin hole of the Raptor and Raptor Reloaded was 5.2mm wide, the Cyborg Beasts was only 4.1mm wide.
The Flexy Hand had the lowest failure strengths of all the designs tested, ranging from 1.6-1.8 kgf (Figure 24). This was due to the live-hinge between the proximal phalange and palm not being rigidly attached by a solid fixator like a metal or 3D printed pin of the previous designs. Rather, it used an interference fit where the live-hinge is press fit into a socket. Although this design choice is structurally weak, it allows the finger to be quickly reattached without the use of tools should the live-hinge come out of its socket. This could easily be done by the wearer or a family member, and eliminates repairs by a specialist.
69


The results of the lateral failure mode test indicated that there was no dependence on strain rate (Figure 27). In the development of this failure mode as a standard test method, the adherence to a strict cross-head speed would not be required for consistent results. Parts from the Raptor hand, which showed high variance in their recorded fracture forces, and parts from the Cyborg Beast, which exhibit low variance, both adhered to this trend. The post-hoc Tukey HSD of the test results showed a statistically significant difference between the mean fracture strengths of the Flexy Hands 5mm/min test, at 1.6 Kgf, and its 500mm/min test, at 1.8 Kgf (Figure 39). However, this was only a 11.1% difference, and given the Flexy Hands low fracture strength at all strain rates, at less than 2 Kgf. This could be considered a statistical anomaly.
The lateral failure mode test developed in this study can be completed quickly, with the 5mm/min test only taking minutes to run, and the lOOOmm/min test only taking seconds. The simplicity of the test set-up makes it well suited to be repeated by other groups, and across a wider range of devices and designs. Establishing a standard failure mode testing method would allow others to contribute their findings to communities such as e-NABLE, and a database of results from more hand designs and print materials can be formed.
Modified SHAP Test. The three designs tested on their ability to perform 25 SHAP tasks showed that power grasp tasks were the most easily completed. This result would be expected given each designs single flexion motion. Tasks requiring extension and tripod both were the next 2 most successful, as the hands were well suited for grasping objects between the thumb and index finger in lateral prehension, a movement
70


achieved in both patterns. The Flexy Hand was successful in these same motions, but achieved them through thumb and index finger opposition, rather than lateral prehension.
Where the hands all consistently failed was in lifting the heaviest abstract object, the sphere. The weight of the sphere overcame the amount of force that could be applied to the hand to successfully hold onto it. The size and shape of the heavy sphere does not allow the user to support the object on a fixed point on the hand, such as the thumb. As explained in the experimental set-up, no objects could be manipulated without the addition of a gripping, high friction surface to the finger-tips. The smooth surface created on the finger-tips during the 3D-printing process is poorly suited for any kind of dexterous or gross manipulations. The Raptor surpassed the Raptor Reloaded in its ability to complete the most tasks, with a 60% completion rate, versus a task completion rate of 44%. This was observed due to the Raptors increased dimensions of the Raptor Reloaded. The wider fingertips and palm allowed for more surface area to be applied to the objects for more secure manipulation. The decreased sized of the revised Raptor Reloaded is a detriment to its overall ability to perform grasping tasks, and was a detriment to its overall usability. This result, along with the Raptors higher failure strength during the lateral failure mode tests indicated that an improvement was not made across the hands design iteration. The Raptor Reloaded did not perform as well as the Raptor in either test. The Flexy Hand consistently failed tasks that required the manipulation of the larger objects of the SHAP tasks. The fixed thumb in a forward adducted position reduced the overall size the hand can be opened compared to the Raptor and Raptor Reloaded, which are designed to be in a naturally abducted posture.
71


The failures experienced during the completion of the SHAP tasks shows that the hands do not have the long-term durability that is an essential requirement for prosthetics. While these failures may have been a nuisance that could be quickly remedied in a workshop setting, they would be a major issue if a user were wearing them day-to-day. The small size of the flexion line requires precise tweezers and grasping tools to effectively tie knots. The complex knots required experience to tie properly, and as demonstrated, did not guarantee a secure fixation. Re-anchoring tension pins into their corresponding screw required a screwdriver, a tool that may not be readily available.
This issue was encountered again when grip-strength testing was attempted. Due to these issues, no reliable or consistent grip strength data could be gathered, as the hands were unable to apply a consistent load without an error arising that would invalidate the test, i.e., knot slippage, tension pin pullout. In a real-world scenario, the wearer would have to forgo functionality of their device until repairs could either be made at home by the family, or if they were incapable of making the repairs themselves, by their specialist, an even bigger inconvenience that would likely lead to device abandonment.
Conclusion
The results of the tensile testing showed that the ABS samples were superior material to the photopolymer ABS substitute RGD525. While RGD525 was stronger, its brittleness leaves it ill-suited for prosthetics, where robustness and ability to withstand everyday wear-and-tear is a necessity. In this study, ABS and RGD525 were compared to 3D-printed composites as well. While the performance of 3D-printed carbon-fiber was significantly greater than the examined, their costs is still a major hurdle to those seeking a low-cost alternative to traditional prosthetics. Two material brands were compared, and
72


printing parameters were controlled during all tests, but there are other parameters of the materials and 3D printing process that were not. The color of the filaments used in the ABS trials was not controlled for. The ambient environment in which the FDM prints were made was not controlled, i.e., ambient temperature, humidity, barometric pressure. Further studies could examine these effects and how they may apply to producing 3D-printed prosthetics. Only two print orientations were examined during tensile testing, flat and vertical. More complex changes to print orientation should be examined in future studies.
Results of the lateral failure mode for a standardized test method of hands showed that strain rate was not a major consideration for the 3D-printed components of prosthetic devices made in ABS. The consistency and accessibility of the results gathered could be readily translated into other modes of failure, such as distant finger joints, wrist joints, or other orientations of the proximal joint. The simplicity of the test set-up makes it easily accessible to others looking to evaluate their own 3D-printed hand designs, or test other designs not examined in this study. The next step in creating a standard would be to partner with either standards designating body, such as ASTM, or a recognized leader in prosthetics research, in either academia or industry, to do further tests on other devices designs that were excluded from this study. The results of this test from a wide array of hands would allow for a consensus to be reached on what values are acceptable for robustness and durability that is suitable for use in daily life of patients. Hands compared against the standard could be given a pass/fail, allowing for redesigns or rejection.
The lack of coordination between independent device designers and medical professionals can be solved if standards that address the concerns of both groups become
73


available. This study lays the groundwork for such a standard to become widespread, and offers a method of testing that can be built upon and remain accessible. Only 4 e-NABLE hand designs were examined in this study, while dozens more are available open-source through various sources. This standard testing method could be applied in further studies, where a database of results could be constructed, giving a comparison of the robustness of all available designs. This could be a source of reliable and verifiable knowledge for clinicians and specialist to decide which hand designs are best suited for their patients.
Assessment of the SHAP tasks with the three selected hand designs showed that current 3D-printed hand designs have a very limited range of motion and postural grip patterns they can achieve, or at least mimic. The failures that occurred to each design during the tasks is even more telling, showing that these designs are not ready for longterm use. While inexpensive and colorful, the hands are awkward to use, cannot support heavy objects, and have many points of failure. In their current state, the designs should be considered little more than toys, and their functionality compares poorly to traditional prosthetic devices. However, they are colorful, inexpensive, and aesthetically pleasing, and could serve as an important aid in helping a limb-deficient child develop socially and prepare them for fitting and use of a traditional prosthetic later in life. This in-itself is an important goal for prosthetics and specialists in treating pediatric amputations, but further work must be done to improve the reliability and robustness of 3D-printed hand designs, if an inexpensive and functional prosthetic for children is to become a reality.
74


REFERENCES
3D-Printable Prosthetic Devices. (2017, 10 5). Retrieved from NIH 3D Print Exchange https://3dprint.nih.gov/collections/prosthetics
Acrylonitrile Butadiene Styrene (ABS) Typical Properties Generic ABS. (2017, August 18). Retrieved from UL Prospector:
https://plastics.ulprospector.com/generics/l/c/t/acrylonitrile-butadiene-styrene-
abs-properties-processing
ASTM International. (2014). Standard Test Methodfor Tensile Properties of Plastics. West Conshohocken: ASTM International.
Beginner's Guide to 3D Printing. (2017, August 18). Retrieved from 3D Insider: http ://3 dinsider. com/3 d-printing-guide/
Biddiss, E. A., & Chau, T. T. (2007). Upper limb prosthesis use and abandonment: A survey of the last 25 years. Prosthetics and Orthotics International, 236-257.
Brusilovsky, D., Gothait, H., Levy, A., & Napadensky, E. (2004). United States Patent No. WO 2004096514 A3.
Davids, J. R., Wagner, L. V., Meyer, L. C., & Blackhurst, D. W. (2006). Prosthetic
Management of Children with Unilateral Congenital Below-Elbow Deficiency. Journal of Bone and Join Surgery, 88(6), 1294-1300.
e-NABLE. (2017, August 17). About us. Retrieved from Enabling the Future: http://enablingthefuture.org/about/
EOS. (2017, August 17). Systems and Equipment for Metal Manufacturing. Retrieved from EOS e-Manufacturing Solutions: https://www.eos.info/systems_solutions/metal
Gothait, H. (1999). United States Patent No. 6,259,962 Bl.
I heart RC Hobby. (2017, April 28). Retrieved from I heart RC Hobby: http://iheartrchobby.com/
ISO 10328:2016. (2017, August 17). Retrieved from International Organization for Standardization: https://www.iso.org/standard/70205.html
ISO 22523:2006. (2017, August 17). Retrieved from International Organization for Standardization: https://www.iso.org/standard/37546.html
Kate, J. t., Smit, G., & Breedveld, P. (2017). 3D-Printed Upper Limb Prosthesis: A Review. Disability and Rehabilitation: Assistive Technology, 300-314.
Krausz, N. E., Rorrer, R. A., & Weir, R. F. (2015). Design and Fabrication of a Six-
Degree-of-Freedom Open Source Hand. IEEE Transactions on Neural Systems and Rehabilitation Engineering.


Krebs, D. E., Edelstein, J. E., & Thornby, M. A. (1991). Prosthetic Management if Children with Limb Deficiencies. Physical Therapy, 77(12), 920-934.
MarkedForge Materials. (2017, August 18). Retrieved from MarkedForge: http s: //m arkforged. com/m ateri al s/
Metcalf, C. D., Woodward, EL, Wright, V., Chappell, P. EL, Burridge, J. EL, & Yule, V.
T. (2008). Changes in Hand Function with Age and Normative. The British Journal of Hand Therapy, 73(3), 79-83.
Murphy, K. (2017, February 17). College students build 3D-printed prosthetic hands for kids. USA Today.
NinjaTek. (2017, September 11). Retrieved from NinjaTek: https://ninjatek.com/
Polylactic Acid (PLA) Typical Properties. (2017, August 18). Retrieved from UL
Prospector: https://plastics.ulprospector.com/generics/34/c/t/polylactic-acid-pla-properties-processing
ProductGoGo. (2017, April 28). Retrieved from ProductGoGo: http://www.productgogo.com/
Resnik, L. (2011). Development and testing of new upper-limb prosthetic devices:
Research designs for usability testing. Journal of Rehabilitation Research and Development, 697-706.
Shetty, P., Menezes, L. T., Tauro, L. F., & Diddigi, K. A. (2013). Amniotic Band Syndrome. Indian Journal of Surgery, 401-402.
Smith, D. G. (2006, January). Limb Loss in Children: Congenital Limb Deficiencies & Acquired Amputations. inMotion, 75(1).
Stratasys. (2017, August 31). Compare Poly Jet Materials. Retrieved from Stratasys.com: http://www.stratasys.com/materials/polyjet/compare-polyjet-materials
Vasluian, E., Bongers, R. M., Reinders-Messelink, H. A., Dijkstra, P. U., & van der Sluis, C. K. (2014). Preliminary study of the Southampton Hand Assessment Procedure for Children and its Reliability. BMC Musculoskeletal Disorders, 199-212.
Ventola, C. L. (2014). Medical Applications for 3D Printing: Current and Projeted Uses. Pharmacy and Theraputics, 704-711.
Zuniga, J., Katsavelis, D., Peck, J., Stollberg, J., Petrykowski, M., Carson, A., &
Fernandez, C. (2015). Cyborg Beast: A Low-cost 3D-printed Prosthetic Hand for Children with Upper-Limb Differences. BMC Research Notes, 8-10.
76


Full Text

PAGE 1

DEVELOPMENT OF A STANDARD OF TESTING AND EVALUATION FOR 3D PRINTED PEDIATRIC UPPER LIMB PROSTHETICS by BRENDAN ROBERT LYLE B.S., Colorado School of Mines, 2014 A thesis submitted to the Faculty of the Graduate School of the University of Colorado in fulfillment of the requirements for the degree of Master of Science Bioengineering Program 2017

PAGE 2

ii This thesis for the Master of Science degree by Brendan Robert Lyle has been approved for the Bioengineering Program by Richard Weir, Chair Cathy Bodine Steven Lammers Date: December 16, 2017

PAGE 3

iii Lyle, Brendan R. (M.S., Bioengineering Program ) Development of a Standard of T esting and Evaluation for 3D Printed Pediatric Upper Limb Prosthetics Thesis directed by Research Associate Professor Richard Weir ABSTRACT The purpose of this study was to identify which common polymers utilized for 3D print ing provide optimum mechanical properties for use in pediatric upper limb prosthetics and to develop a standard method of failure mode testing which can be applied across a wide range of devices Selected devices were also tested for their ability to complete existing standard usability tasks. Results of mechanical testing showed that ABS filament provided the most robust material properties, and that print orientation showed no significant effect on ultimate tensile strength. The s tandard failure mode test developed showed consistent results across all designs t ested and was quickly repeatable and accessible to those unfamiliar with mechanical testing. Tasks requiring a tripod or lateral prehensile grip on lightweight objects were most successful, while large and heavy objects and spherical prehensile grip were the least. These findings will be useful in the development of a standard of testing and evaluation for 3D printed upper limb prosthetics so that they can be better understood by the medical and clinical community, identifying their strengths and weaknesses in a straightforward manner. The form and content of th is abstract are approved. I recommend its publication Approved: Richard Weir.

PAGE 4

iv ACKNOWLEDGEMENTS The author would like to acknowledge the following individuals for their cont ributions and assistance in the completion of this study: Dr. Richard Weir Dr. Ca thy Bodine Dr. Steven Lammers Dr. Levin Sliker Dr. Bradford Smith Stephen Huddle

PAGE 5

v TABLE OF CONTENTS I. INTRODUCTION ................................ ................................ ................................ ........... 1 LITERATURE REVIEW ................................ ................................ ................................ 2 Part 1: Background on Childhood Amputations and Pediatric Prosthetics ................. 2 Amputation Causes. ................................ ................................ ................................ 2 Pediatric Amputee Statistics ................................ ................................ .................... 6 Age for Fitting and Outcome Effect ................................ ................................ ........ 7 Prosthetic Requirem ents and Types ................................ ................................ ......... 9 Outcomes of Prosthetic Use and Design Evaluation ................................ ............. 15 Part 2: 3D printing and the Current State of 3D Printed Prosthetics ......................... 20 3D Printing Hardware ................................ ................................ ............................ 20 FDM Plastics ................................ ................................ ................................ .......... 24 3D Printed Prosthetics ................................ ................................ ............................ 25 SPECIFIC AIMS ................................ ................................ ................................ ........... 27 I I. EXPERIMENT AND RESULTS ................................ ................................ ................ 28 Experimental Set up ................................ ................................ ................................ .. 28 Construction of 3D printed hands. ................................ ................................ ......... 28 Inclusion and Exclusion Criteria for 3D Printed Hand Designs. ........................... 29 Printers and Software. ................................ ................................ ............................ 32 Selection of Plastics. ................................ ................................ .............................. 34 Tensile Testing. ................................ ................................ ................................ ...... 35 Lateral Testing for Standard Development. ................................ ........................... 38 Usability Modified SHAP test. ................................ ................................ ........... 42 Results ................................ ................................ ................................ ....................... 44 Tensile Testing. ................................ ................................ ................................ ...... 44 Lateral Failure Mode Testing. ................................ ................................ ................ 48 SHAP Task Results. ................................ ................................ ............................... 57 Discussion ................................ ................................ ................................ .................. 59 Construction Observations. ................................ ................................ .................... 59 Tensile Test Results. ................................ ................................ .............................. 65 Lateral Failure Mode Testing. ................................ ................................ ................ 67 Modified SHAP Test. ................................ ................................ ............................. 70

PAGE 6

vi Conclusion ................................ ................................ ................................ ................. 72 REFERENCES ................................ ................................ ................................ ................. 75

PAGE 7

vii LIST OF TABLES Table 1 Student t test p values of tensile tested plastics. Upper left Comparing plastic types in Flat orientation. Lower left Comparing plastic types in Vertical print orientation. Right Comparing Flat vs Vertical Print orientations for each tested plastic. ................................ ................................ ................................ ................................ ........... 48 Table 2 SHAP Task results from each tested hand on pass/fail basis and primary prehensile pattern utilized for each task. ................................ ................................ ........... 58 Table 3 Overall Completion rates from each hand design based on primary prehensile pattern ................................ ................................ ................................ ............................... 59

PAGE 8

viii L I S T OF FIGURES Figure 1 Prosthetics for child effected by thalidomide Science Museum of London, licensed under CC BY SA 2.0. ................................ ................................ ................................ ......... 4 2 Hand of infant affected by Amniotic Band Syndrome (Image in public domain). ....... 5 3 Time to failure of attending patients based on age of initial fitting (Davids et al. 2006, pg.1296). Used with permissions by Wolters Kluwer Health, Inc. ................................ .... 9 4 Passive mitten (Krebs et. al, 1991) Used with permissions by Oxford University Press ................................ ................................ ................................ ................................ ........... 10 5 Child sized voluntary opening hook (Krebs et al. 1991, pg.927). Used with permissions by Oxford University Press. ................................ ................................ .......... 11 6 Child's myoelectric hand and glove (Krebs et al. 1991, pg.926). Used with permissi ons by Oxford University Press. ................................ ................................ .......... 13 7 a) Raptor Hand Render by Jeremy Simon is licensed under CC BY 3.0 b) Raptor Reloaded c) Fle xy Hand d) Cyborg Beast Render by Creighton Labs is licensed under CC BY NC ................................ ................................ ................................ .............................. 29 8 Phoenix Hand v2 by EnableCommunityFoundation i s licensed under CC BY NC ... 31 9 Failed Pheonix Hand v2 print due to insufficient surface area for proper filament adhesion. ................................ ................................ ................................ ........................... 32 10 ASTM D638 Type 1 Tensile Specimen W=13mm; L=57mm; WO=19mm; LO=165mm; G=50mm; D=115mm; R=76mm; T=3.2mm. W c and WO are de fined in ASTM D638 for alternate specimen types or specific materials not utilized in this study (ASTM International, 2014). ................................ ................................ ............................ 36 11 Carbon Fiber (left) and Reinforced Nylon (right) Type 1 Tensile Specimens. ......... 36

PAGE 9

ix 12 a) Flat print orientation b) Vertical print orientation. ................................ ................ 37 13 a) Raptor test jig b) Raptor Reloaded test jib c) Cyborg Beast test jig d) Flexy Hand test jig. ................................ ................................ ................................ ............................... 39 14 a) Test jig mounting bracket b) Lateral Failure Mode tensile hook. ......................... 40 15 Laterail Failure Mode Test Setup (post break). ................................ ......................... 41 16 Raptor Hand with Lee Tippi Micro Gel fingertips. ................................ ................... 43 17 Camry EH101 hand dynamometer. ................................ ................................ ........... 43 18 Stress Strain curves of the Flat and Vertical Generic ABS tensile specimens. ......... 44 19 Stress Strain curves of Flat and Vertical Hatchbox ABS tensile specimens ............. 45 20 Stress Strain curves of Flat and Vertical RGD525 tensile specimens. ...................... 45 21 Mean Ultimate Tensile Strengths from all sp ecimens and print orientations with standard error. ................................ ................................ ................................ ................... 46 22 Mean Ultimate Tensile Strengths from all specimens and print orientations with expected literature values and composite samples. ................................ ........................... 47 23 Cyborg Beast Lateral Failure Mode Test results ................................ ....................... 49 24 Flexy Hand Lateral Failure Mode Test results ................................ .......................... 49 25 Raptor Lateral Failure Mode Test results ................................ ................................ .. 50 26 Raptor Reloaded Lateral Failure Mode Test results ................................ .................. 50 27 Mean Failure Strengths of the tested hand designs versus strain rate with standard deviation. ................................ ................................ ................................ ........................... 51 28 Raptor ANOVA Table ................................ ................................ ............................... 52 29 Raptor ANOVA table statistics ................................ ................................ ................. 52 30 Raptor Tukey HSD results ................................ ................................ ......................... 53

PAGE 10

x 31 Raptor Reloaded ANOVA Table ................................ ................................ ............... 53 32 Raptor Reloaded ANOVA Table statistics ................................ ................................ 54 33 Raptor Reloaded Tukey HSD results ................................ ................................ ......... 54 34 Cyborg Beast ANOVA table ................................ ................................ ..................... 55 35 Cyborg Beast ANOVA table statistics ................................ ................................ ...... 55 36 Cyborg Beast Tukey HSD results ................................ ................................ .............. 56 37 Flexy Hand ANOVA table ................................ ................................ ........................ 56 38 Flexy Hand ANOVA table statistics ................................ ................................ ......... 57 39 Flexy Hand Tukey HSD results ................................ ................................ ................. 57 40 Render of Flexy Hand finger build plate. ................................ ................................ .. 62 41 Print failure of Flexy Hand printer build plate due to improperly constructed CAD file. ................................ ................................ ................................ ................................ .... 63 42 Example of tension pin splitting common among all designs tested. ........................ 64 43 Gauntlet failures of Flexy Hand (a) and Raptor (b). Also note the tension pin block split in the Flexy Hand gauntlet from tightening of flexion lines. ................................ .... 64 44 Proximal Phalange joint failure experienced in all lateral mode failure tests. From left to right: Cyborg Beast, Raptor Reloaded, Raptor. ................................ ...................... 69

PAGE 11

1 I INTRODUCTION The rise of 3D printing technology in the last decade has caused a surge of interest in a wide array of fields of medicine. The use of 3D printed materials for custom made, inexpensive, and quickly produced prosthetic devices has gained world wide attention through media and popular culture (Murphy, 2017) Wh ile these reports provide som e insight into the design of hands and how they may be used, there is a distinct lack of information into their benefits over existing devices. Reviews of the technology and apparent benefits exist, but they tend to make many generalizations about the capability of 3D printed devices, while not providing first hand validation (Kate, Smit, & Breedveld, 2017) What is not being addressed is the lack of understanding among clinicians, occupational the rapists, patients, and their families as to how these devices should best be used or what considerations need to be made for the devices to be successful and improve the quality of life for the user. Designs for 3D Printed Prosthetic hands are available online from many sources, but e NABLE is perhaps the most well known, providing resources for creators, makers, health professionals, and patients to connect and receive help in getting a device printed to meet their needs (e NABLE 2017) E NABLE has provided designs used in previous studies of 3D printed prosthetics, but these are generally literature reviews of available designs, their basic functions, and recommendations for fitting and sizing (Zuni ga, et al., 2015) Many studies done with traditional prosthetics show that abandonment rates of users are commonly in excess of 50% if the device does not meet their needs or fails to improve their functionality, and these rates climb even higher fo r children (Biddiss & Chau, 2007) What the literature lacks is an engineering analysis 3D printed prosthetic

PAGE 12

2 designs, and effective guidelines written with clinicians and pediatric prosthetic specialists in mind for what cons iderations must be made for fabricating prosthetics using 3D printers and how best they can be utilized. This project will address these issues and provide the basis on which production standards and future evaluations can be done and answer the questio n of what do production standards for 3D printed prosthetics need to address, and how should they be conducted. The results of these tests and examinations can be utilized in the future in academic and industry publications to formulate standards as well as guidelines to make inexpensive, functional, and appealing prosthetics a viable choice for childhood amputees and their families, as well as provide an analysis of existing designs and 3D printing technology to further improve the development cycle of ne w devices moving forward. LITERATURE REVIEW Part 1: Bac kground on Childhood Amputations and Pediatric P rosthetics Amputation Causes. Knowing the cause of an amputation and understanding how they are categorized is important in establishing what treatment options individual patients should receive, and how their needs should be addressed. There are two classifications under which all amputations fall, congenital limb deficiency and acquired amputations. Congenital limb deficiency is a broad term that encompasses any birth every 1,000 births. Deficiencies are more likely to occur in upper limbs than in lower limbs, and even more unlikely for a deficiency to be present in both the lower or upper extremities. Total congenital limb loss is also rare (Smith, 2006) Smith referred to all

PAGE 13

3 other deficiencies as acquired limb deficiency, encompassing loss of limb ranging from vascular disorder, cancer, traumatic injury, etc. Smith wrote from the perspective of giving information to the parents of children born with con genital amputations, who are largely responsible for deciding what care their child receives and what treatments and prosthetic devices they receive. They are also responsible for following up with treatments provided by clinicians and occupational therap inMotion an online publication of the Amputee Coalition provided background information on congenital amputations that is easily condit ion. The concerns of the parents of childhood amputees are a key aspect to recognize in the development of prosthetics as well, as they ultimately must assist their child in learning how to use and adapt to their devices. Congenital limb deficiencies are caused by a variety of factors. Exterior causes may come from exposure of the mother to hazardous chemical substances, such as the infamous outcome of thalidomide prescription to pregnant mothers in the 1950s (Krebs, Edelstein, & T hornby, 1991) The example of a disabled by the effects of thalidomide in

PAGE 14

4 Figure 1 shows the extreme severity of limb deficiencies that occurred due to the drugs administration. prosthetic harness designed for a child Figure 1 Prosthetics for child effected by thalidomide Science Museum of London, licensed under CC BY SA 2.0. Another exterior cause is early amnion rupture, which can result in amniotic band syndrome (Davids, Wagner, Meyer, & Blackhurst, 2006) Amniotic Band Syndrome occurs when the limbs of the developing fetus are entrapped by stands of amniotic tissues, which inhibit development and cause a wide array of deletions and deformations

PAGE 15

5 (Shetty, Menezes, Tauro, & Diddigi, 2013) An example of an hand effected by Amniotic Band Syndrome is shown in Figure 2. Figure 2 Hand of infan t affected by Amniotic Band Syndrome (Image in public domain). In most cases, however, congenital limb deficiency is an isolated condition not associated with the musculoskeletal system. These deficiencies are likely the result of mutations or vascular compromise to the apical ectodermal ridge (D avids, Wagner, Meyer, & Blackhurst, 2006) The exact cause of these mutations is unknown. Both Krebs et al. and Davids et al. discussed the background of congenital amputations in their publications on childhood prosthetics, despite their difference in subject matter. Krebs et al. cited the use of thalidomide in prompting the growth of interest in the field of pediatric prosthetics, which at the time had been the most studied cause of childhood amputations. Davids et al. published their long term o utcome study 15 years after Krebs et al. published their overview of pediatric prosthetic devices, and

PAGE 16

6 there is notably a shift in focus towards genetic mutation and idiopathic occurrences. The increased knowledge in the field of genetics between the publication of these two papers shows the increased understanding that researchers have gained in addressing what causes of childhood amputations are most significant. Shetty et al.( 2013) further expands on the notion established by Davids et al that Amniotic Band Syndrome can be caused by early amnion rupture, again focusing in on greater detail with a specific case study to better un derstand and establish the causes of congenital limb deficiencies. Pediatric Amputee Statistics Due to the wide range of amputation causes, types, and methods of reporting by health facilities, there is little comprehensive data on the population of ch ildhood amputees in the United States (Krebs, Edelstein, & Thornby, 1991) Krebs et al. provides insight into the epidemiologic factors that lead to amputations in children, and sites his own study into the demographics of chi ldhood amputees. The study from 1984 was taken by 4,105 children being treated at specialty clinics for limb deficiencies. From the data received by Krebs from these clinics, it was estimated that of all cases, 17% had been first identified in the past y ear, with a further three fifths being due to congenital limb reductions (Krebs, Edelstein, & Thornby, 1991) This concluded that there are 5,525 new cases of childhood amputations to be expected each year, with 3,315 of these being congenital. Citing this same study, Davids et al. stated that unilateral congenital below elbow deficiencies are the most common limb deficiency managed in North America, thus the need for prosthetic devices that address these deficiencies is the g reatest (Davids, Wagner, Meyer, & Blackhurst, 2006) Both Davids et al. and Krebs et al. noted the lack of available data from clinical sources on childhood amputees. Both Krebs et al. and Davids et al. cited a separate stud y

PAGE 17

7 done by Krebs in 1984 on the etiology of childhood amputees, making it one of the few definitive sources on patient statistics, despite its age. However, given the endemic nature of childhood amputations, it is safe to assume that the instance rate of c hildhood amputees in the US has not changed significantly. Both groups also recognized from Krebs study that the most abundant group of childhood amputees suffer from unilateral congenital below elbow deficiencies. It follows that demand for devices that address patients with these deficiencies is the highest. Age for Fitting and Outcome Effect The age at which a child is fitted for a prosthetic is an important consideration for clinicians. Krebs et al. identified the need to fit a prosthetic as early as possible, with ages of less than 3 being the most widely reported (Krebs, Edelstein, & Thornby, 1991) Davids et al. performed a long term study from 1954 to 2004 with 260 c hildhood patients exhibiting upper limb below elbow deficiencies. The goal of this study was to determine the outcomes of long term prosthetic management in children with unilateral congenital below elbow deficiency (Davids, Wagner Meyer, & Blackhurst, 2006) A successful outcome was defined as a child and parents who continued attending the limb deficiency clinic and claimed during follow ups that a prosthetic had been worn outside the clinical setting for any period of time (Davids, Wagner, Meyer, & Blackhurst, 2006) An unsuccessful outcome was defined as a child and parent who no longer followed up at the clinic or reported that the child did not wear a prosthesis (Davi ds, Wagner, Meyer, & Blackhurst, 2006) The study found that fitting a child below the age of 3 provided the best long term outcome for long term prosthetic use, but also recognized that fitting before the age of one had little impact on the outcome of long term prosthetic use. The most important factor in a successful

PAGE 18

8 long term outcome for prosthetic use was intensive training under the direction of the clinics occupational therapists, and utilizing a variety of prosthetic designs over the rowth. This progression of designs began with a passive device, fitted once the child had learned to walk. Providing positive results and interest from the child and parents, the child was then moved into a body powered device between the ages of two to f our, and finally myoelectric devices of increasing complexity (Davids, Wagner, Meyer, & Blackhurst, 2006) The age for fitting a child in a prosthetic device is addressed by both Davids et al. and Krebs et al. While Krebs e t al. did not address a minimum age for fitting, it recognized that the optimal age for fitting is before 3 years, when the child has become old enough to understand instruction from occupational therapists in a clinical setting, and their parents outside of the clinic. Davids et al. similarly concluded that before the age of 3 gives the best outcome for long term success, with the stipulation of avoiding fitting before the age of one. Davids et al. based on age of initial fitting is shown in Figure 3.

PAGE 19

9 Figure 3 Time to failure of attending patients ba sed on age of initial fitting ( Davids et al. 2006, pg.1296 ). Used with permissions by Wolters Kluwer Health, Inc. Davids et al. more on age of fitting than Krebs et al. but their findings motor skills. The emphasis in both papers on this information shows the importance of fi tting children with devices, rather than waiting until they are more developed. Thus, there is a clear need for prosthetics specifically designed to meet the needs of children. Prosthetic Requir ements and T ypes Pediatric prosthetics require special considerations to meet the needs of children and their parents. Not only must the prosthetics be smaller, but the physical and psychological growth of the child must be considered (Krebs, Ede lstein, & Thornby, 1991) The needs of the parents also factor into the prosthetic prescription; some families may want a functional device fitted on their child as soon as possible, while others may want an aesthetic passive device to disguise the am putation. (Krebs, Edelstein, & Thornby, 1991) Financial considerations must also be considered by the parents.

PAGE 20

10 P assive devices are the simplest of all prosthetics. They consist of rubber or plastic forms modelled to look like the missing appendage in both shape and color. A p assive mitten sized for a child is shown in Figure 4. While they are unable to provide function that replaces the deficient limb, infants fitted with passive devices have been observed using the device to stabilize their movements, such as when transitioni ng from sitting to standing (Krebs, Edelstein, & Thornby, 1991) However, Davids et al. suggested that a passive prosthetic should not be fitting to the child until it has learned to walk, for fear that it may interfere with d evelopmental motor skills. Body driven systems are the next progression in prosthetic design. The most common and most diverse body driven terminal device is the hook. An example of a body powered hook sized for a child is shown in Figure 5. Figure 4 Passive mitten (Krebs et. al, 1991) Used with permissions by Oxford University Press

PAGE 21

11 Figure 5 Child sized voluntary opening hook ( Krebs et al. 1991, pg.927 ). Used with permissions by Oxford University Press. This allows for bimanual prehension, an important motor skill in early childhood development (Krebs, Edelstein, & Thornby, 1991) The open close motion of the hook is commonly controlled by a fitted shoulder harness, to which a cable extending from the hook is run up the arm and anchored to the opposing shoulder. Adduction of the shoulder transmits movement through the cable to open or close the attached device. Modifications of the hook exist that more closely resemble the form and ergonomics of the human hand, providing a more natural grasping pattern and aesthetics. These devices can be covered with a cosmetic glove to further disguis e the amputation. However, the gloves can discolor and tear, requiring replacement (Krebs, Edelstein, & Thornby, 1991) The most complex prosthetics are myoelectric units. These units are driven by electronic motors to control hook or hand to open and close. Electronic Electromyogram (EMG) sensors are placed on the skin above corresponding muscles groups to detect electrical signals when the muscle is vo luntarily flexed (Krausz, Rorrer, & Weir, 2015)

PAGE 22

12 This signal correspond s to a specific motion on the device. For example, flexion of the bicep will close a hand, while flexion of the tricep will open it. This is a common arr angement, as the agonist antagonist arrangement is easy for the user to understand and control. Depending on the level of the amputation, there may be a limited number of motor control sites available (Krausz, Rorrer, & Weir, 2015) This leaves many designs restricted to single degrees of freedom, which can be limited for adults, but in children, can provide a useful training tool. While the most functional of prosthetic devices, it is uncommon for children to be fitted with myoelectric devices. While battery powered devices eliminate the need for harnesses and cables, the unit is usually heavier and more fragile (Krebs, Edelstein, & Thornby, 1991) Additionally, the child may have to wear the ba ttery pack on their body or waist if the space inside the proximal device and the terminal device is not large enough to house it internally. The added weight and discomfort of such devices, along with their fragility and complex operation make them less suited for children, though infants as young as 18 months have been fitted (Krebs, Edelstein, & Thornby, 1991) Like body driven hooks and hands, myoelectric devices can be fitted with aesthetic sleeves to better disguise the example of a child sized myoelectric hand is shown in Figure 6.

PAGE 23

13 Figure 6 Child 's myoelectric hand and glove ( Krebs et al. 1991, pg.926 ). Used with permissions by Oxford University Press. All of these systems are classified as terminal devices, which replace a missing hand. Prosthetics also require proximal devices to mount the terminal device to the residual limb. Behind the terminal device is commonly a wrist unit, which provides passive pronation and supination. Wrist units also exist that can provide palmar flexion (Krebs, Edelstein, & Thornby, 1991) Past the wrist unit are the remaining proximal units. On body driven systems, this consists of the harness and cable system, while on myoelectric units, this will include signal processors, motor control units, and battery packs. Attachment of the proximal device to the body is done through flexible sockets with a rigid frame custom fit to the patient (Krebs, Edelstein, & Thornby, 1991) The size form of this unit is uniquely depending on the level of amputation for the patient. As the child grows, progre ssively larger devices will need to be exchanged. Commercial terminal devices and wr ist units typically come in a range of sizes that can be modularly swapped out (Krebs, Edelstein, & Thornby, 1991) Sockets and rigid bodies must be resized or rebuilt to fit the patients growing body. The type of prostheti c selected for the child is dependent on many factors, as pointed out in the literature. One factor that was only briefly mentioned by Krebs et al. is

PAGE 24

14 the factor of cost. In their study on low cost prosthetic hands, Zuniga et al. quoted the price of a body powered system at $4,000USD $20,000USD (Zuniga, et al., 2015) Krausz et al. estimated the price of state of the art myoelectric hands at $30,000USD, and that figure does not include the additional cost of proximal devices. Unlike adults, childhood prosthetics must be refitted and replaced to keep up in size and function as the child grows. The cumulative costs of fitting a child with such an expensive device every one to two years is high. There are many more options available for adult amputees, and for childhood amputees when funds are unlimited but there is a distinct lack of low cost options that provide a similar level of function. Davids et al. studied the time to abandonment in their pati ents based on the type of device used. Passive devices were abandoned the quickest, while myoelectric devices w ere abandoned the least often, with body drive devices between. The time to failure based on the patients studied by Davids et al. who wore var ies types of prosthetic devices is show in Figure 7. Figure 7 Time to Failure of Patien ts based on prosthetic design ( Davids et al. 2006, pg.1297 ). Used with permissions by Wolters Kluwer Health, Inc.

PAGE 25

15 This shows that the use of functional prosthetics are more desirable to patients than ones that just disguise a limb deficiency. Outcomes of Prosthetic Use and Design Evaluation It is the goal of physical or occupational therapists working with amputee s to find the best outcome for their patients, whether that includes the use of a prosthetic or no device at all. Davids et al. noted that the children in their study could perform most tasks without the use of a prosthetic, and if the device did not enha typically be abandoned. (Davids, Wagner, Meyer, & Blackhurst, 2006) Over the course of the 50 year study, 49% of Davids et al. re reported when the child and parents regularly attended the clinic for therapy and training in their devices (Davids, Wagner, Meyer, & Blackhurst, 2006) This suggests that the type of device used is less important than how well familiarized the patient is with it, and the level of support in its use provided Davids et al. conceded that their study may over or under estimate the number of successful patient outcomes, as their pass/fail criteria was very broad. In a similar study on device abandonment, Biddiss & Chau. looked at 25 years of prosthetic user data across 200 articles to determine what dissa tisfactions users had with devices that led to the ir abandonment. They divided their findings based on the type of device: passive, body powered, and myoelectric. Biddiss & Chau found that passive devices had t he lowest abandonment rate, as low as 6%, but with some reports as high as 100% (Biddiss & Chau, 20 07) They theorize the user s relative satisfaction was due to the simplistic nature of passive devices and their cosmetic appeal. The concerns with passive devices included were wear temperature, glove malfunctions, weight, wear on

PAGE 26

16 clothing, and ir ritation from straps. Rates of abandonment for body powered devices were reported between 16% and 66%, with lower rates of rejection among adults than children (Biddiss & Chau, 2007) ovement, aw kward use, difficult cleaning maintenance, weight, low grip strength, and high energy expenditure as well as appearance, as the most prominent reasons for abandonment (Biddiss & Chau, 2007, p. 241) Biddiss & Chau also noted the prominent popularity of body powered systems despite more advanced systems being available, due to their functionality, durability, lower weight, visibility of objects being handled, and accessibility of designs. Rates of abandonment for myoelec tic designs ranged from 0% to 75%. Biddiss & Chau hypothesized that the low abandonment rate of myoelectric devices stem from advancing designs, availability, and the growing culture of technology t s of weight, appearance, and discomfort that they share d with other types of devices, myoelectric devices were reported to have the additional issue s of battery life and charging, motor and sensor reliability, and an unsatisfactory range of motion in the f inger and wrist units (Biddiss & Chau, 2007) While improvements on design and functionality have been prevalent in the literature, and rel iability. Even the best device will be rejected if the user cannot use it easily. Usability has become a prominent buzzword in the field of prosthetics and assistive technology. Dr. Linda Resnik examined usability and how it relates to the development a nd testing of upper limb devices. Resnik began with the concept of device

PAGE 27

17 ease of use of device user interfaces (Resnik, 2011, p. 6 98) In their work, Resnik cited the ISO definition of usability as the most relevant to medical devices: The extent to which a product can be used by specified user to achieve specified goals with effectiveness, efficiency, and satisfaction in a sp ecified context of use (Resnik, 2011, p. 698) Resnik recognized that previously there was little consensus on the most important aspects that need to be considered for evaluating the performance of prosthetics. The big g est issue in prosthetics is there are so many concerns that must be addressed in order to create a successful outcome Resnik clearly addressed that for prosthetic usability to be improved, designs cannot be created in a vacuum. The device must meet the needs, and that studies must be done with feedback on device performance and usability methods and case studies, which cannot be easily repeated and verified. It i s clear from created and adopted uniformly for future developments in prosthetics to be successful. One such system for quantifying the success in the use of a prosthetic d evice is the Southampton Hand Assessment Procedure (SHAP). The SHAP test measures the due to its thorough overview of the six prehensile grips (spherical, tripod, p ower, lateral, tip, extension) and a simple to understand general functionality score (Vasluian, Bongers, Reinders Messelink, Dijkstra, & van der Sluis, 2014) The six prehensile patterns are illustrated in Figure 8.

PAGE 28

18 Figure 8 Six Prehensile Patte rns examined by the SHAP test (Metcalf, et al., 2008) Used with permissions by SAGE Publications, Ltd. The SHAP test involves 26 tasks, 12 where the patient manipulate s abstract objects, and 14 where the p atient performs an activity of daily living (ADL). The time to complete the task is recorded in seconds, and using a z score transform of each time and the Euclidean distance, the six prehensile patterns and an overall index of function (IOF) is calculate d. These scores range from 1 to 100, where 100 is normal functionality. These scores are compared to a predetermined norm, so a score of over 100 is possible if the patient scores above the norm. These norm values are unavailable due to intellectual pro perty rights of the commercial party that distributes the SHAP test kit (Vasluian, Bongers, Reinders Messelink, Dijkstra, & van der Sluis, 2014) The study by Davids et al. recognized the need for data on long term prosthetic use, starting in early childhood. Davids et al. recognized that a 49% abandonment rate of subjects in the study is significantly high, and that one way to decrease user abandonment is to increase the v ariety of devices available to patients, to best suit a wide array of needs (Davids, Wagner, Meyer, & Blackhurst, 2006) The study performed by Vasluian et al. attempted to develop a new set of norm values so that the SHAP tes t may be applied to

PAGE 29

19 children, in what they called the SHAP C (Vasluian, Bongers, Reinders Messelink, Dijkstra, & van der Sluis, 2014) Their study also modified the objects and protocols in the SHAP test itself. The size of s ome of the abstract and ADL objects was reduced or modified for easier manipulation by a child. Other tasks were modified so the completion criteria would be easier, such as the test assessor assisting the child in initially grasping certain objects. Bot h these studies showed that unique criteria must be considered not only when developing prosthetic devices for child ren, but also in how performance is assessed, whether it be directly assessing the device, or determining success by long term adoption. A t the time of this writing, there are no recognized standards on evaluating the mechanical capabilities of upper limb prosthetics. Two ISO standards exist for evaluating lower limb prosthetic. ISO 10328:2016 defines structural testing of lower limb prost heses (ISO 10328:2016, 2017) ISO 22675:2016 defines testing for ankle foot devices and foot units (ISO 22523:2006, 2017) Neither of these standards is publicly available, with their contents be hind a paywall of $205.67USD each. This cost is a large hurdle for an independent prosthetic designer or researcher, and the increasing interest in 3D printed prosthetics will require standards to ensure that devices meet the expectations of health provid ers and families, for both upper and lower limb devices. The development of a standard through a body such as ASTM, which coordinates with research institutes to make their database available to researchers, or with a group such as e NABLE, who will be di scussed later, would help reduce costs of independent prosthetic development and promote standardization between designs.

PAGE 30

20 Part 2: 3D printing and the Current State of 3D Printed P rosthetics 3D Printing Hardware 3D printing, or additive manufacturing, is any method by which objects are created by fusing or depositing layers of material sequentially until the piece is completed (Ventola, 2014) 3D printing applications in medicine has a thorough overview of 3D printing te chnology. They recognized the three main categories of printers that are commercially available, though many methods of additive manufacturing have been developed since the tec hnologies The first method of 3D printing discussed by Ventola was Selective Laser Sintering (SLS). SLS printing utilizes a laser passed over a powder bed of the desired material. This laser heats the powder bed in a precise s pot until the particles become fused together. This process is done in the shape of the object, layer by layer until completion (Ventola, 2014) SLS is capable of printing metals, ceramics, and plastics. The resolution of th e piece is controlled by the precision of the laser and the size of the powder. Figure 9 illustrates the inner working of an SLS printer, and demonstrates the sintering occurring at the particulate level.

PAGE 31

21 Figure 9 Cross section of SLS p rint bed and sintering process Wikimedia licensed under CC BY SA Hospital Colorado Research Institute. It has the capability to print in aluminum, Cobalt Chromium, titanium, and several steel alloys, with a print volume of 228mm x 228mm x 190 mm (EOS, 2017) The next method of 3D printing Ventola discussed is Fused Deposition Modeling (FDM). FDM printers are the most common and least expensive 3D pri nters available. They operate by extruding a continuous strand of heated plastic filament out of a print head, similar to that of a tradition al ink printer, onto a print bed. Like SLS, this print head moves in the shape of the desired object, layer by la yer, until completed (Ventola, 2014) FDM printers allow for greater versatility in their selection of polymers available, and modularity in designs, such as multiple print heads. In Figure 10, a typical FDM 3D printer is ill ustrated.

PAGE 32

22 Figure 10 Schematic representation of the 3D printing technique known as Fused Filament Fabrication; a filament a) of plastic material is feeded through a heated moving head b) that melts and extrudes it depositing it, layer after layer, in the desired shape c) A moving platform e) lowers after each layer is deposited. For this kind of technology additional vertical support structures d) are need ed to sustain overhanging parts. By Paolo Cignoni, licensed under CC BY SA The FDM used during this study was a FlashForge Creator Pro, made available through the University of Colorado Department of Bioengineering. The last method of 3D printing highlighted by Ventola is Thermal Inkjet Printing (TIJ). TIJ is not a method of 3 D printing used for structural components, but rather for tissue reconstruction and bioprinting. Ventola explained the process as follows : thermal, electromagnetic, or piezoelectric technology

PAGE 33

23 to depos materials) onto a substrate according to digital instructions. In inkjet printing, droplet deposition is usually done by using heat or mechanical compression to eject the ink drops. In TIJ printers, heating t he printhead creates small air bubbles that collapse, creating pressure pulses that eject ink drops from nozzles in volumes as small as 10 to 150 picoliters. Droplet size can be varied by adjusting the applied temperature gradient, pulse frequency, and ink viscosity (p 705). The limited material volume of TIJ printing makes it unsuitable for large 3D printed structures, and thus TIJ was not utilized in this study. There an additional method of printing utilized in this study not discussed by Ventola, and that is the proprietary photopolymer printing method utilized by Objet Connex printers. These printers deposit a continuous sheet of photopolymer resin onto the print bed in the defined shape of the object, layer by layer (United S tates Patent No. 6,259,962 B1, 1999) This photopolymer is cured by a UV light after the completion of each layer to provide a rigid surface for the next layer to be deposited on. The properties of the photopolymer resins vary based on proprietary c ompositions, but characteristics of the available resins range from hard plastics, ductile plastics, and rubber like elastomers (United States Patent No. WO 2004096514 A3, 2004) This method of printing w as available at the ti me of Vent

PAGE 34

24 The versatility of materials Objet printers offers makes their use in 3D printed prosthetics appealing, as parts made multiple mate rial types can be printed along side each other, while EOS printers are limited to one material type The material versatility of FDM printers are limited by the number of print heads, typically only one or two. The limited number of printers described by Ventola demonstrates that the literature available to rese archers and professionals in the medical field may not cover always cover all available technology. Further work must be done to analyze all avail able methods of 3D printing in the medical world, not only in the field of prosthetics, but all areas of medi cine. FDM Plastics The types of materials use in 3D printers varies widely based on the printer being used, especially in FDM printers, where unlike EOS SLS printer or Objet Photopolymer printers, the user is not restricted by proprietary materials tha t must be sources from the manufacturer. The two most common plastics used in filament for FDM printers are ABS and PLA. ABS, or Acrylonitrile Butadiene Styrene is a common thermoplastic derived from petroleum byproducts. ABS is considered both strong a nd ductile, and is used frequently in manufacturing for packaging, toys, and any number of injected molded plastic products (Beginner's Guide to 3D Printing, 2017) As an amorphous solid, it has a glass transition temperature of approximately 105C but no true melting point, making it well suited for 3D printing (Acrylonitrile Butadiene Styrene (ABS) Typical Properties Generic ABS, 2017) PLA, or Polylactic Acid, is a thermoplastic derived from pla nt byproducts (Beginner's Guide to 3D Printing, 2017) Like ABS, it is strong but also brittle, making it less suited for use in devices that require durability or resistance to fracture. Unlike ABS, PLA has an organized crys tal structure,

PAGE 35

25 giving it a glass transition temperature of approximately 57C and a melting point of approximately 161C (Polylactic Acid (PLA) Typical Properties, 2017) Like ABS, this is an ideal temperature range for 3D printing. More exotic materials have also become commercially available for use in FDM printers. The MarkedForge Mark Two is capable of printing continuous strands of carbon fiber, nylon, Kevlar, and fiberglass impregnated within a polymer filament typic al of traditional FDM printers (MarkedForge Materials, 2017) They are advertised as having mechanical properties above that of usual 3D printed thermoplastics, and even above 6061 T6 Aluminum. This claim makes them very appe aling for use in prosthetics, where both high strength and light weight are necessary requirements for successful devices. 3D Printed Prosthetics The lowering costs and availability of 3D printers has made them an appealing choice in manufacturing prosth etics. The widespread use of CAD modeling software among both professional and amateur designers has also led to a wide array of prosthetic designs available online. A widely recognized source of 3D printed prosthetic designs is e NABLE. Their organizat ions mission statement describes themselves as: [A community of] teachers, students, engineers, scientists, medical professionals, tinkerers, designers, parents, children, scout troops, artists, philanthropists, dreamers, coders, makers and every day peopl e who just want to make a (e NABLE, 2017)

PAGE 36

26 e NABLE acts as a hub where designers can share their files for prosthetic hands that are purpose made to be created using 3D pri nters. The website also offers families of childhood amputees the ability to contact and connect with owners of 3D printers who wish to produce prosthetic devices for those looking for alternative solutions for their iting, e NABLE provides files and resources for 7 wrist actuated designs, also known as tenodesis action, and two elbow actuated designs. T 3D Print Exchange also hosts these same designs, and ted prosthetics is curated by e NABLE volunteers (3D Printable Prosthetic Devices, 2017) Very little clinical evaluation of the 3D printed devices available from e NABLE has been conducted. Zuniga et al. examined one design available through e NABLE, the Cyborg Beast. In their study, Zuniga et al. proposed a method for sizing and fitting the hand appropriately for a user remotely, so that the hand may be delivered to the user ready to use, without furthe r adjustment needed by the family. They found that their method of remote fitting was successful, but that there is a need for further studies that examine functionality, validity, durability, benefits, and rejection rates of 3D printed hands (Zuniga, et al., 2015) A review of 3D printed upper limb prosthetics by Kate et al. included 5 e NABLE designs out of a total of 58 devices. While thorough, this review only examines the mechanical and kinematic specifications provided by the designers. The study goes into great detail on aspects such as the range of motion of individu al joints, number of joints, degrees of freedom, extension/flexion methods, type of control etc. This data may not be useful to the end user, as it is not pre sented in a manner that helps a clinician or

PAGE 37

27 occupational therapist in selecting a device, and no ne of it is verified through actual testing and physical evaluation of the hands. While Zuniga et al. provided a sound method for fitting, and Kate et al. provided detailed technical data on designs, none of it provided information on how the devices perf orm ed how patients should use them, what sort of durability is expected, or if patients will even ac cept them. This is the gap in kn owledge on 3D printed hands this study will address. SPECIFIC AIMS From the literature review, it is clear that 3D print ing can help solve many issues with current prosthetic technology, mainly cost, accessibility of devices, and functionality with aesthetic appeal. However, to date there is little work done on how 3D printing can be st be applied to pediatric prosthetics. There is no uniformity in information on which printers and materials are best used in prosthetics. There is no easily accessible information for specialists for which of the dozens of open source 3D printable hands available provide the best features or designs that would best suit their patients. Finally, there is no easily accessible guide for specialists on how the available open source designs should be constructed, starting from the printing stage, to post possessing, and final assembly. This stud y will address these gaps by providing the preliminary steps needed to create standards which can be utilized by specialist in the field of pediatric prosthetics to best serve their patients and answering exactly how 3D printed prosthetics can best be used The following specific aims will be addressed : Specific Aim 1: Document and evaluate printing and construction considerations of currently available open source 3D printed upper limb prosthetics.

PAGE 38

28 Specific Aim 2: Determine the Suitability of Structural Plastics used currently in 3D printing for use in upper limb prosthetics. Specific Aim 3: Evaluate hand designs for usability by the patient. II EXPERIMENT AND RESULTS Experimental Set up Construction of 3D printed hands. With regards to Specific Aim 1 the CAD files for the hands evaluated in this study were all obtained through e NABLE. The Raptor Hand, Raptor Reloaded, Cyborg Beast, and Flexy hand were chosen for evaluation at various steps of the study. Renders of these hands ar e shown in Figure 7

PAGE 39

29 Figure 7 a) Raptor Hand Render by Jeremy Simon is licensed under CC BY 3.0 b) Raptor Reloaded c) Flexy Hand d) Cyborg Beast Render by Creighton Labs is licensed under CC BY NC Inclusion and Exclusion Criteria for 3D Printed Hand Designs. The Raptor and Raptor Reloaded were chosen to examine two iterations of the same basic design for improvement. The Flexy Hand was chosen due to its use of live hinges, rather than pinned hinges, and its natural prehensile thumb design. The Cyborg Beast was selected due to its use occurrence in the literature as well as its use of non 3D printed parts. For all testing, all components of the Flexy Hand were produced at 70% scaling to represent

PAGE 40

30 the dimensions of a device intended for a child. All other hands were left at their default scaling, as t hey were suitably sized for use by children at 100% scaling. When applicable, all components for the hands were printed for a right hand user. For consistency in testing, no left hand designs or v ar iants were utilized during the study. Designs available outside of e NABLE were excluded in order to keep the scope of the experiment concise and ensure the hands evaluated will be accessible for future studies. All files are available in .stl format, which is accessible by the majority of CAD software packages, including Solidworks, AutoCAD, and Microsoft 3D Builder. All hands selected required additional m anufactured hardware to be built, none of the selected designs could be built solely from 3D printed materials. All hands required wood screws for the anchoring and adjustment of tension pins that secure the tension lines and kept the flexion lines taught. Braided high strength f ishing line was used in all hands to act as tension lines for the generation of finger flexion motion. Braided elastic string was used for finger extension when the force generated by the wearer in wrist flexion was ceased. The Flexy Hand did not require the use of elastic string for finger flexion, as the polyurethane live hinges return the hand to an open position when force generated by the wearer s wrist is ceased. The Cyborg Beast requires the addition of a metal Chicago bolt to pin the proximal fin ger pieces to the palm, rather than a printed pin (Raptor, Raptor Reloaded) or interferenc e socket (Flexy Hand) that was used in the other hand designs. A fifth design, the Phoenix Hand v2, was initially included in the evaluation process (Figure 8 ).

PAGE 41

31 F igure 8 Phoenix Hand v2 by EnableCommunityFoundation is licensed under CC BY NC The Phoenix Hand was chosen due to its similar design and construction to the Raptor hands, with mechanical hinges and pins, but features an adducted and opposed thumb similar to the Flexy Hand. The design also featured enclosed tracks that the flexion lines needed to be fed through to reach their tension pin anchors. However, the addition of supports during the 3D printing process filled these tracks and subsequently could not be removed from such a small enclosed space. Attempts to reorient the part on the build platform to alleviate the need for support in the flexion line tracks resulted in

PAGE 42

32 failure, as a piece as large as the palm could not be supported by the two sma ll, round wrist hinges (Figure 9 ). Figure 9 Failed Pheonix Hand v2 print due to insufficient surface area for proper filament adhesion. The palm of the Flexy Hand uses the same method of guiding its flexion wires, but the rear of the palm is flat, allowing it to be printed upright easily, thus solving the issue of removin g support from within the flexion line tracks. Given the large amount of material that would be needed in the trial and error process needed to solve this issue, the Phoenix Hand was eliminated from the study. Printers and Software. For the completion of Specific Aim 1 and 2, two FDM printers were utilized to construct all components for mechanical testing purposes were by FlashForge USA, City of Industry CA This printer is a dual print head design with a heated build platform, and enclosed build chamber for ambient temperature stability. One printer was modified with a Dual Flexion flexible filament extruder by Diabase Engineering, Longmont CO. This allowed for printing of

PAGE 43

33 NinjaFlex polyurethane filamen t to be used in construction of the live hinge components of the Flexy Hand. The control s oftware for the FDM printers used at the beginning of the study was MakerBot Desktop. This package is available free of charge from 3D printer manufacturer Makerbo t Brooklyn NY The software allows for the import of .stl files and converts them into G code, which is read by the printer as a series of sequential x y z coordinates to guide the print heads into the proper shape of the desired object. The software pa ckage also allows for control of print head and print bed temperatures depending on the selected plastic, as well as options to adjust infill percent, part scaling, place support material, rafts, and adjust the quality of the print. Test prints with Maker Bot Desktop resulted in components that were significantly undersized from their expected dimensions. The software also could not produce rafts of consistent quality that allowed the plastic to adhere properly to the print bed. Attempts to address these issues by changing print parameters within the software w ere unsuccessful. Control software for the FDM printers was then changed to Simplify3D by Simplify3D, Inc., Blue Ash OH Simplify3D is a license based 3D printer control software package with more features than MakerBot Desktop. The switch to Simplify3D corrected all issues in test prints that came from the use of MakerBot Desktop. All photopolymer printing was done on a n Objet Conne x350. This printer includes it s own proprietary control softwar e that adjusts temperatures and UV exposures automatically depending on the selected materials. CAD files are imported and the user can control scaling, placement on the print bed, and build quality.

PAGE 44

34 Selection of Plastics. With regards to Specific Aim 2, ABS was selected for examination on all hand parts and material test specimens. PLA was excluded due to its brittleness and tendency to fracture under stress rather than deform. Prosthetics in practical use are dynamic devices and there is the expecta tion from the wearer that the device will be robust and able to stand up to everyday wear and tear, as well as minor bumps and drops. PLA does not allow for this expectation to be met, thus its exclusion. Two brands of ABS were used to examine if variatio n in filament manufacturer would need to be a consideration of clinicians producing 3D ected due to its competitive price, and ease of availability from multiple retailers. These are parameters that would be a significant consideration by clinicians when selecting which filaments to purchase for the construction of 3D printed devices. Gene ric ABS and Hatchbox ABS were used to construct tensile specimens. Only Hatchbox ABS was used to construct full hands and parts for usability testing, due to the limited quantity of Generic ABS available in the lab space. NinjaFlex by NinjaTek was used t o construct the elastomer live hinges for the All Generic ABS samples were printed with the extruder at 230C and 100C build platform. All Hatchbox samples and device parts were printed with the extruder at 226C and 106C build platform. The extruder speed for both ABS examples was 3600 mm/min. All prints done in ABS were done with the print area enclosed, with the printer door closed and the lid on, as to provide a stabl e ambient temperature. All NinjaFlex parts were printed with a 230C extruder and a 40C print bed, and with a print speed of

PAGE 45

35 1500mm/min, per the manufactures recommendations (NinjaTek, 2017) Prints in NinjaFlex were done with the printer door clos ed but with the lid removed, to promote faster cooling. All prints utilized the default raft generated in Simplify3D. Supports on all prints were added using the automated support generator feature in Simpify3D. All parts mad e on the Objet Connex350 were built with RGD525. Objet the available Objet Connex350 printer, thus a comparison between actual ABS filament is warranted (Stratasys, 2017) This photopolymer was used to construct tensile specimens as well as hands for usability testing. For a basic comparison of available composite 3D printed materials relative to ABS and RGD525, single tensile specimens were sourced from outside vendors constructed in carbon fiber reinforced and Nylon reinforced plastics from a Marked Forge 3D printer. The 3D printed carbon fiber testing specimen was sourced from I Heart RC Hobby, Salt Lake City, UT, and the 3D printed Nylon reinforced specimen was sourced from ProductGoGo, Littleton, CO (I heart RC Hobby, 2017; ProductGoGo, 2017) Tensile Testing. To complete Specific Aim 2, tensile testing on both brands of ABS, RGD525, and the carbon fiber and Nylon reinforced samples wa s done in accordance with ASTM D638 (ASTM International, 2014) All tensile specimens were printed to meet the specified Type 1 dimensions for rigid plastics. The dimensions of the ASTM Type 1 tensil e specimen are shown in Fig ure 10

PAGE 46

36 Figure 10 ASTM D638 Type 1 Tensile Specimen W=13mm; L=57mm; WO=19mm; LO=165mm; G=50mm; D=115mm; R=76mm; T=3.2mm. W c and WO are defined in ASTM D638 for alternate specimen types or specific materials not utilized in this study (ASTM International, 2014). Tensile specimens made in ABS were printed at 100% infill to achieve the highest possible strength. Tensile specimen default infill setting. The single carbon fiber specimen was received with the printed specimen was received with Nylo n fibers oriented along the perimeter of the specimen. The as received carbon fiber and nylon tensile specimens are shown in Figure 11 Figure 11 Carbon Fiber (left) and Reinforced Nylon (right) Type 1 Tensile Specimens.

PAGE 47

37 Two versions of ABS and RGD 25 tensile specimens were produced to examine the effect a parts orientation on the print bed ha on the resulting mechanical properties. ed a 90 orientation is shown in Figure 12 a, and the Vertical Print orientation is shown in Figure 12 b. Figure 12 a) Flat print orientation b) Vertical prin t orientation. The first set of vertical and flat RGD525 specimens and all Generic ABS tensile samples were tested on an MTS Insight tensile frame. All remaining tensile specimens were tested on a Mark 10 ESM1500 tensile frame. This change was made to in crease the volume of tests that could be performed, as the Mark 10 ESM1500 was more readily accessible to the researcher. Readings from load cells on both machines were measured and recorded in newtons, and displacement was measured and recoded in millime ters. All tensile specimens were pulled at a speed of 5mm/min. Results of tensile tests were

PAGE 48

38 rejected if the specimen fractured outside of the designated gauge length, per ASTM D638, with the exception of the Carbon fiber and Nylon reinforced samples, as only one was available. Lateral Testing for Standard Development. To standardize mechanical testing for 3D printed prosthetics and give a measure of usability outlined in Specific Aim 3, a universal method of testing hands for relative durability in an easily reproducib le and common mode of failure was needed. A lateral force on the fifth digit was seen to be a typical location for high and rapid loading, such as if the wearer were to stand up from a seated position and support their body weight on the hand. This motion was replicated using a Mark 10 ESM1500 tensile frame with custom designed mounting brackets for mounting and imparting force on test sections of each hand design. The hands tested in this mode of failure were the Raptor, Raptor Reloade d, Cyborg Beast, and Flexy Hand. The palm of each hand was modified into a test jig using Microsoft View 3D to reduce the part down to a quartered section containing the mounting position for the fifth digit. Renderings of these test jigs are shown in Fi gure 13

PAGE 49

39 Figure 13 a) Raptor test jig b) Raptor Reloaded test jib c) Cyborg Beast test jig d) Flexy Hand test jig. Attempts to modify the CAD files for the selected designs were originally made in SolidWorks 2015, but the high polygon count of these parts was not easily convertible by the software from .stl to the proprietary .sldprt file format. The Flexy Hand palm .stl file for example, has a surface comprised of over 80,000 polygons. This caused software crashes and hardwar e freezes regularly, and no progress could be made. The switch to Microsoft View 3D solved this issue, as the software does not require the conversion of the .stl to a proprietary file format for edits to be made, and the part is rendered as a wire frame model rather than a solid body, reducing the computing power needed to render the models.

PAGE 50

40 Mounting brackets for holding the modified palms and fifth finger were assemble d in the proper position to simulate a lateral load and a hook to pull the digits that could be held by the tensile frame grips were designed in SolidWorks 2015. Renders of the mounting bracket and t ensile hook are show in Figure 14 Figure 14 a) Test jig mounting bracket b) Lateral Failure Mode tensile hook. Both the mounting bracket and the tensile hook were printed on an EOS M270 metal 3D printer in maraging steel for strength properties that would far exceed the plastics be ing tested, as to avoid failures from the testing equipment. Support removal and su rface finishing of the steel components included a combination of hand held electric grinders, end milling, and sand blasting. Both testing components have a large, thin area for inserting into the tensile test grips of the Mark 10 ESM1500. This provided sufficient surface area for t he grips to latch on to so the testing components did not come loose during the lateral testing procedure. Mounting holes were drill ed in t he modified palms and mounting bracket so they could be quickly swapped in and out and were secured by bolts. The hook for

PAGE 51

41 pulling the fifth digit was placed at the joint between the proximal phalange and intermediate phalange for each trial. The complet e te sting setup is shown in Figure 15 Figure 15 Laterail Failure Mode Test Setup (post break). Readings from load cells on both machines were measured and recorded in newtons, and displacement was measured and recoded in millimeters. The speed at which lateral testing occurred was done at 5mm/min, 100mm/min, 500mm/min, and 1000mm/min. Speeds were chosen to represent the lowest loading speed, as designated by ASTM D638, and increased upward through 2 orders of magnitude. Varying the loading rate was done to determine if parts constructed through 3D printing exhibit a strain rate dependence, a commo n mechanical property of plastics. All parts were printed

PAGE 52

42 at 40% infill to represent typical 3D printer parameters, which are usually selected to reduce the overall amount of material used, and thus reduce cost. Usability Modified SHAP test. For the completion of Specific Aim 3, a modified SHAP test was performed on the Raptor, Raptor Reloaded, and Flexy Hand to determine what tasks and what prehensile patterns were easily completed by the 3D printed devices. SHAP tasks were not timed, as t he purpose of the examination was to determine which tasks were achievable, not how quickly they could be achieved. As shown by Vasluian et al, the SHAP test is inherently flawed when performed by children, since all normative values used in its score calcu lations are intended for adults. Thus, all SHAP scores would be invalid. Hands were manipulated by a non limb deficient user in the seated position defined by SHAP parameters. Each task was evaluated under a pass/fail criteria if the task could be compl eted under the parameters of the SHAP test. Care was taken not to orient the device in such a way that would be considered un natural, and only in a manner in which a typical user could achieve i.e. beyond 90 perpendicular to the users seated position a nd the test area, or inverted. All SHAP tasks carton would have proved a hazard to the electrical equipment in the researchers shared lab space. Initial attempts at pe rforming tasks revealed that no objects could be manipulated, as the smooth plastic of the fingers did not create enough friction for objects to be securely held. All hands were fitted with Lee Tippi Micro Gel Fingertip Grips for final testing so that obj ects would not slip as easily while being manipulated ( Figure 16 ).

PAGE 53

43 Figure 16 Raptor Hand with Lee Tippi Micro Gel fingertips. Grip strength tests was attempted during the study, using a Camry EH101 hand dynamometer and the 3 hands selected for use in the SHAP task evaluations. The dynamometer is shown in Figure 17 Figure 17 Camry EH101 hand dynamometer.

PAGE 54

44 It was discovered that the dynamometer could not be fit into the hand in any position that allowed for consistent and repeated grip strength to be measured. Flexing the hands against the dynamometer handle regularly caused the flexion line knots to untie and loosen, and the tension pins to pull out from the anchoring screws. Due to these repeated failures, data could not be consistently collected. This test was not pursued further in this study. Results Tensile Testing. Engineering stress and strain of the Generic ABS in both flat and vertical print orientations are sho wn in Figure 18 Each line represents one specimen trial(n=3). Figure 18 Stress Strain curves of the Flat and Vertical Generic ABS tensile specimens.

PAGE 55

45 Engi neering stress and strain of the Hatchbox ABS in both print orientations are sh own in Figure 19 (n=6). Figure 19 Stress Strain curves of Flat and Vertical Hatchbox ABS tensile specimens Engineering stress and strain of RGD525 in both print o rientations are show in Figure 20 (n=7). Figure 20 Stress Strain curves of Flat and Vertical RGD525 tensile specimens.

PAGE 56

46 The mean ultimate tensile strengths(UTS) from all trials, from all three materials, in both print ori entations is show in Figure 21. Figure 21 Mean Ultimate Tensile Strengths from all specimens and print orientations with standard error.

PAGE 57

47 A comparison of the mean UTS of the tested plastics with the e xpected UTS values from the literature as well as the results of the single sample 3D printed composites are shown in Figure 22 Figure 22 Mean Ultimate Tensile Strengths from all specimens and print orientations with expected literature values and composite samples. Statistical analysis of the tensile test results was performed using 2 group Student t tests between each set of test result s: first, comparing the effect of materials within orientation groups, and then comparing the effect of orientation within material groups. The resulti ng p values are shown in Table 1

PAGE 58

48 Table 1 Student t test p values of tensile t ested plastics. Upper left Comparing plastic types in Flat orientation. Lower left Comparing plastic types in Vertical print orientation. Right Comparing Flat vs Vertical Print orientations for each tested plastic. p value p value FLAT FLAT v. VERTICAL Generic v. Hatch 0.2596 Generic ABS 0.0116 RGD525 v. Generic 4.81E 05 Hatchbox ABS 0.0123 RGD525 v. Hatch 5.58E 06 Photo. 525 0.471 VERTICAL Generic v. Hatch 3.32E 04 RGD525 v. Generic 2.60E 03 RGD525 v. Hatch 3.77E 04 No statistical significance in resulting UTS is shown between the Generic ABS and Hatchbox ABS in the flat orientation (p>0.05). All other material comparisons when controlling for print orientation on resulting UTS show statistical significance (p<0.05). Print orientation is shown to be statistically significant in resulting UTS for booth Generic ABS and Hatchbox ABS (p<0.05). Print orientation is shown not to be statistically significant in resulting UTS for RGD525 (p>0.05). Lateral Failure Mode Test ing. Force deflection plots of the Cyborg Beast hand in the purposed Lateral Failure mode test at 5mm/min, 100mm/min, 500mm/min, and 1000mm/min st rain rates are shown in Figure 23 Results of the same test on the Flexy Hand, Raptor, and Raptor Reloaded, are shown in Figure 24, 25, and 26 respectively.

PAGE 59

49 Figure 23 Cyborg Beast Lateral Failure Mode Test results Figure 24 Flexy Hand Lateral Failure Mode Test results

PAGE 60

50 Figure 25 Raptor Lateral Failure Mode Test results Figure 26 Raptor Reloaded Lateral Failure Mode Test results

PAGE 61

51 The mean failure strength of each hand at the four strain rates is shown in Figure 27 Figure 27 Mean Failure Strengths of the tested hand designs versus strain rate with standard deviation. Statistical analysis of the lateral failure mode test began with a one way ANOVA of the mean failure strength of each hand design at each strain rate. The ANO VA Tables statistics, and Tukey HSD results for each hand are shown in Figure s 28 39 A post hoc Tukey Honest Significant Difference test was then performed on the ANOVA results 0 5 10 15 20 25 30 35 40 45 50 5 100 500 1000 Fracture Strength ( Kgf ) Strain Rate (mm/min) Mean Fracture Strength in Lateral Pull Flexy Hand Cyborg Beast Raptor Reloaded Raptor

PAGE 62

52 from each hand design to determine if the mean failure strengths between eac h strain rate were significantly different. Figure 28 Raptor ANOVA Table Figure 29 Raptor ANOVA table statistics

PAGE 63

53 Figure 30 Raptor Tukey HSD results Figure 31 Raptor Reloaded ANOVA Table

PAGE 64

54 Figure 32 Raptor Reloaded ANOVA Table statistics Figure 33 Raptor Reloaded Tukey HSD results

PAGE 65

55 Figure 34 Cyborg Beast ANOVA table Figure 35 Cyborg Beast ANOVA table statistics

PAGE 66

56 Figure 36 Cyborg Beast Tukey HSD results Figure 37 Flexy Hand ANOVA table

PAGE 67

57 Figure 38 Flexy Hand ANOVA table statistics Figure 39 Flexy Hand Tukey HSD results P values from the resulting ANOVA on each hand design demonstrated there was not a statistically significant difference between the resulting mean failure strength and strain rate (p>0.05). The post hoc Tukey HSD determined that there was no mean failure strength significantly different regardless of strain rate in all cases exce pt one: the Flexy Hand, displayed a statistically significant difference in means between the failure strength at strain rates of 5mm/min and 500mm/min. SHAP Task Results. The pass/fail results of each hand evaluated in the completion of the 2 5 SHAP tas ks is shown in Table 2

PAGE 68

58 Table 2 SHAP Task results from each tested hand on pass/fail basis and primary prehensile pattern utilized for each task. Task Prehensile Pattern Flexy Hand Raptor Raptor Reloaded Wood Extension Extension pass pass pass Metal Extension pass pass fail Page Turning fail fail fail Carton Pouring fail pass pass Lifting Light Object fail pass pass Wood Lateral Lateral pass fail fail Metal Lateral pass fail fail Button Board fail fail fail Lifting Tray pass pass pass Rotate Key fail fail fail Wood Cylinder Power pass pass pass Metal Cylinder pass pass pass Glass Jug Pouring fail pass fail Door Handle pass pass pass Wood Sphere Spherical pass pass pass Metal Sphere fail fail fail Jar Lid fail fail fail Wood Tip Tip pass pass fail Metal Tip pass pass fail Pick up Coins fail pass pass Open/Close Zip fail fail fail Wood Tripod Tripod pass pass pass Metal Tripod pass pass pass Food Cutting fail fail fail Rotate Screw fail fail fail Results were further broken down into how many tasks, based on the primary prehensile pattern, each hand completed. A percentage value of the total number tasks Task Completion ese results are shown in Table 3

PAGE 69

59 Table 3 Overall Completion rates from each hand design based on primary prehensile pattern Spherical Tripod Power Lateral Tip Extension Totals Task Completion Rate Flexy 1 2 3 3 2 2 13 52% Raptor 1 2 4 1 3 4 15 60% Raptor Reloaded 1 2 3 1 1 3 11 44% During the course of the SHAP task evaluations, all hands suffered from mechanical failures. The Flexy Hand had the 5 th digit tension pin pull out from the anchoring screw, and the 4 th digits flexion line knot came undone. The Raptor and Raptor Reloaded both experienced the se same failures during the test. The test was stopped for repairs to be made on the hands, and res umed once the hands and been fixed. Discussion Construction Observations. The use of 3D printers for quickly manufactured, and inexpensive prosthetics is possible, but there are many considerations that were recognized in this study that are not found in existing literature. The set up and maintenance of common FDM 3D printers themselves is a non trivial requirement for any clinic that would want to produce 3D printed devices for their patients. In the course of this experiment, build plate platters needed to be replaced, and print head assemblies were upgraded to meet new mate rial requirements. Filament mis feeding was also a regular occurrence during filament swaps, and each material replacement called for a trial and error phase where both print head and build plate temperatures needed to be discovered for optimal part adhes ion and build quality with each material examined. All these parameters are printer dependent, and no two 3D printers behave identically.

PAGE 70

60 Diagnosing these problems required familiarity not only with 3D printing technology, but with each unique printer. and printers in a clinical setting would result in poor part quality and printer down time as problems persist. Trained staff to maintain and operate 3D printers would be a requirement for 3D printed prosthetics to become comm onplace. The software that controls the parameters and feeds instructions to the printer itself is complex and requires exp erience to be able to use it to its fullest extent and pr oduce the highest quality print Parameters commonly modified during thi s study include d support type, infil type, infil density, extruder and build platform temperature, inclusion or exclusion of raf ts, and print head speed. Fine tuning these parameters must be done by trial and error, and are dependent on the material being used. Changing materials requires these parameters be changed as well. For example, Hatchbox ABS is advertised to require a n extruder temperature in a range of 210C 240C, and a build platform temperature range of 55C 85C. Likewise, NinjaFlex design ates an extruder temperature range of 225C 235C, and a build platform temperature of 40C. Trial and error during this studied showed that an extruder temperature of 226C and build plate temperature of 106C was ideal for Hatchbox ABS, and 230C extrude r temperature and 40C build platform temperature for NinjaFlex. Simplify3D allows for material profiles to be created by the user and quickly interchanged, but these pr ofiles must be made through a trial and error process. Complications with the Makerb ot Desktop software that led t o the switch to Simplify3D showed that available software packages were not innately reliable and their accuracy must be evaluated by the user. The unintended part scaling that was discovered

PAGE 71

61 during this study may not have been found by a less experienced user until parts had already been delivered, such as to a clinician and patient for final fitting. At that point, the problem would have to be diagnosed and the process restarted, delaying th e patient from receiving the care they expect from the clinician. Zenga et al. scaling or significant errors in the measurements could affect the function or fitting of the 3D printed p function may be. If attention by the user is not taken, it is possible for a part to be scaled below the resolution of the printer in use, resulting in a zero dimension feature t hat will fail to print properly. Alternatively, a part may be scaled down so small that robust features at 100% scaling may no longer have structural integrity. The issue of scaling is one that may not be readily apparent to prescribers of 3D printed dev ices. Errors in the files provided through e discovered during this study. The individual phalanges for the construction of the Flexy s pre assembled into one file so that the user does not need to go through the task of counting and scaling each individual phalange for all five fingers. A CAD rendering o f this part is shown in Figure 40

PAGE 72

62 Figure 40 Render of Flexy Hand finger build plate. However, this file contained an error in which the pieces were not zeroed to be flat on the build platform. The thin geometry of the part in the orientation designated in the file resulted in little surface area for adhesio n to the underlying support to fill the space between the part and the build platform. This resulted in numerous attempts where the printer ended up printing in air, where the underlying part had failed to adhere properly and the extruder continued to tra vel upward. The resulting failed print of the Flexy Hand f inger plate is shown in Figure 41

PAGE 73

63 Figure 41 Print failure of Flexy Hand printer build plate due to improperly constructed CAD file. Only after this error was discovere d and corrected could the Flexy Hand be properly constructed. An inexperienced user of the software and in printing these devices maybe not recognize this error until after multiple failed prints have occurred, and maybe not have access to proper CAD soft ware to edit and correct the error in the e NABLE file. Under designed features are prevalent in all hand designs examined. Common among all designs were the use of tension pins for supporting the braided fishing that acts as flexors tendons. The pins from each design have walls too thin to properly serve as thread anchors for the designated screws. The Raptor hand had a tensioner pin wall thickness of o

PAGE 74

64 sized scaling used in this study. As a result, pins splitting when screws were tightened for proper flexor tension resulted reg u larly. This is shown in Figure 42 Figure 42 Example of tension pin splitting common among all designs tested. Another common design feature shared among all the assembled hands was the wrist gauntlet, which provides cut out s for strapping and attachment to the wearers arm. During the course of the study, the gauntlets for both the Flexy Hand and Raptor Hand fractured through no fault but regular wear and use. These are shown in F igure 43 Figure 43 Gauntlet failures of Flexy Hand (a) and Raptor (b). Also note the tension pin block split in the Flexy Hand gauntlet from tightening of flexion lines.

PAGE 75

65 The thin features of these components need to be enhanced in order to provi de a robust and durable device for the patient, otherwise repeated failures may quickly lead to device abandonment. Tensile Test Results. While the difference in UTS for the flat and vertically printed ABS tensile specimens was statistically significant, the practical difference was small. The mean UTS for the flat Hatchbox ABS and vertical Hatchbox ABS was 29.5 MPa and 34.5 MPa, respectively. This is only a 14.6% difference in mean UTS between Hatchbox ABS print orientations. The mean UTS for the flat Generic ABS and vertical ABS was 26.5 MPa and 30.4 MPa, respectively. This is only a 12.8% difference in the mean UTS between Generic ABS print orientations. There was no significant statistical between Hatchbox ABS or Generic ABS when printed in the flat orientation. While the vertical orientation was statistically significant, the practical difference again was small, with a difference of only 11.8%. These results suggested that contrary to the literature, print orientation was not a significant factor in the strength of ABS parts made for 3D printed prosthetics. The effect of print orientation was shown to be significantly insignificant in the specimens printed in RGD525. The mean UTS for the flat RGD525 and vertical RGD 525 was 49.4 MPa and 52.1 MPa, respectively. This is only a 5.2% difference in mean UTS. In the application of 3D printed pediatric prosthetics, which are created with the intent of being inexpensive, and easily replaceable, this small difference in mean UTS between the exami ned print orientations indicated this does not need to be a major consideration.

PAGE 76

66 The standard deviation of the flat and vertical RGD525 specimens was 4.7 MPa and 8.4 MPa, respectively, which was higher than he standard deviations of both Hatchbox ABS(Flat 3.9 MPa, Vertical 1.0 MPa) and the Generic ABS(Flat 1.5 MPa, Vertical 0.4 MPa). This demonstrated t he resulting mechanical properties of devices created using Objet photopolymer printers will not have reliable mechanical proper ties as compared to those printed in ABS from an FDM printer. The stress strain curves of the two tested material types show ed a distinct difference in modes of failure. While both the ABS samples and RGD525 samples exhibited similar elongations of 12% 22%, their failure modes differ. The distinct region of plastic deformation past the UTS shown in both the ABS sample results indicated higher ductility than RGD525, which show almost no region of plastic deformation, and thus brittle failure ( Figures 18, 19, 20 ). Despite the higher UTS seen in RGD525, its wider range of expected mechanical properties and tendency to fail in a brittle manor makes it less suited for 3D printed prosthetic devices than ABS. While ABS had a lower UTS, it is more ductile, and thus any loading beyond its UTS will not result in immediate and complete fracture of a part. This robustness makes ABS the desired material. The comparison of the tested materials to the material properties found in the literature show ed that expected results for both the ABS and RGD525 were achieved. Comparing the two tested materials to the single trial composite samples showed that Nylon re enforced samples were not necessarily stronger than ABS or RGD525, with a UTS of only 65.0 MPa. This was only 19.9% higher than the vertical RGD525 samples, the strongest result observed in this test. However, the tested carbon fiber sample displayed strengths on an order of magnitude higher than the tested plastics, at a UTS of

PAGE 77

67 224.1 MPa ( Figure 22 ). The increased cost of the Marked Forge printers capable of producing parts in carbon fiber goes against the idea that 3D printed prosthetics are an inexpensive alterna tive to traditional prosthetics. A MarkedForge Mark Two printer with composite pri nting capability costs $13,499, and a spool of their proprietary carbon fiber filament is $447 for a 150cm 3 build volume (MarkedForge Materials, 2017) Currently, the overall cost of producing 3D printed composite prosthetics surpasses that of traditional devices. However, if the technology were to become less expensive and more widely available, they would be the superior choice over either ABS or RGD5 25. Lateral Failure Mode Testing. The results of the lateral failure mode test developed for this study are presented as the force applied to the test pieces, in kilograms force, versus how far the finger deflected before fracture, in millimeters. Thes e values were chosen over the standard stress and strain values of pascals and mm/mm in order for them to be more clearly understood by amputee clinicians and specialists who are less familiar with engineering principles. Kilograms force was chosen over p ascals, as it is a more accessible measure of mass relative to the force of gravity, rather than pressure. The SI unit of force in similar tests is newtons. However, in keeping with the premise that this test is to be a standard for those unfamiliar with engineering principles, the more commonly used unit of kilograms force was chosen. In a wider application, this could be further localized into imperial units of pounds force. Furthermore, a unit of force was maintained rather than pressure, as it broad ens the accessibility of the failure mode test to many different device designs without having to know the cross sectional area of various components of the pieces being tested, which would be required for the accurate calculation of applied pressure. In the case of the hands tested, the complex geometries

PAGE 78

68 of the proximal phalanges, the metacarpal joints, and their fixation pins would make this a non trivial process. Strain, which is often considered unitless, is a confusing measure to those unfamiliar wi th engineering principles. Deflection in millimeters provides an fracture. Again, this also requires no previous calculation of gauge length, which when testing comple x shapes, is non trivial to derive. In wider applications, the measurement of deflection could be presented in imperial units of inches or fractions of inches. The Raptor showed the highest consistent failure strengths, 16.2 24.1 kgf ( Figure 25 ), of al l the tested designs, with the Raptor Reloaded being the second highest, at 11.3 16.7 kgf ( Figure 26 ). This is due to the Raptors much larger and robust design, whereas the Raptor Reloaded has been slimmed down and dimensions reduced for a more aesthetica lly pleasing form. The Cyborg Beast strengths ranged from 5.1 5.5 kgf (Figure 23), which performed marginally b etter than the Flexy Hand. The inclusion of the Cyborg Beast in this study was to examine how hand designs with sourced metal hardware performe d compared to fu lly 3D printed designs. What was shown was that the addition of metal fixture proved no better at improving mechanical strength than their 3D printed counterparts. The use of metal hardware goes with the assumption that the weak point of 3D printed hands are the fixture pins. This study showed that not to be the case. Rather, all tests run in the lateral failure mode had failure occur at the rear pin hole of the proximal phalange, and no failures of the pin itself occurred on either the R ap tor or Raptor Reloaded (Figure 44 ).

PAGE 79

69 Figure 44 Proximal Phalange joint failure experienced in all lateral mode failure tests. From left to right: Cyborg Beast, Raptor Reloaded, Raptor. was only 3.0mm wide. The Cyborg Beast experienced much lower failure strengths due to its dimensionally thinner components. The pin hole of the Raptor and Raptor Reloaded w The Flexy Hand had the lowest failure strengths of all the designs tested, ranging from 1.6 1.8 kgf ( Figure 24 ). This wa s due to the live hinge between the proximal phalange and palm not being rigid ly attached by a solid fixator like a metal or 3D printed pin of the pr evious designs. Rather, it used an interference fit where the live hinge is press fit into a socket. Although this design choice is structurally weak, it allows the finger to be quick ly reattached without the use of tools should the live hinge come out of its socket. This could easily be done by the wearer or a fam ily member, and eliminates repairs by a specialist.

PAGE 80

70 The results of the lateral failure mode test indicate d that there w as no dependence on strain rate ( Figure 27 ). In the development of this failure mode as a standard test method, the adherence to a strict cross head speed would not be required for consistent results. Parts from the Raptor hand, which show ed high varianc e in their recorded fracture forces, and parts from the Cyborg Beast, which exhibit low variance, both adhere d to this trend. The post hoc Tukey HSD of the test results showed a statistically significant difference between the mean fracture strengths of t test, at 1.6 Kgf, and its 500mm/min test, at 1.8 Kgf ( Figure 39 ). However, this was only a 11.1 % difference, and given the Flexy Hands low fracture strength at all s train rates, at less than 2 Kgf. This could be considered a statistical anomaly. The lateral failure mode test developed in this study can be completed quickly, with the 5mm/min test only taking minutes to run and the 1000mm/min test only taking seconds. The simplicity of the test set up makes it well suited to be repeated by other groups, and across a wider range of devices and designs. Establishing a standard failure mode testing method would allow others to contribute their findings to communities such as e NABLE, and a database of results from more hand designs and print materials can be formed. Modified SHAP Test. The three designs tested on their ability to perform 25 SHAP tasks showed that power grasp tasks were the most easily completed. This result would be expected given each design and tripod both were the next 2 m ost successful, as the hands were well suited for grasping objects between the thumb and index finger in lateral prehension a movement

PAGE 81

71 achieved in both patterns. The Fl exy Hand was successful in these same motions, but achieved them through thumb and index finger opposition, rather than lateral prehension. Where the hands all consistently failed was in lift ing the heaviest abstract object, the sphere. The weight of the sphere overcame the amount of force that c ould be applied to the hand to successfully hold onto it The size and shape of the heavy sphere does not allow the user to support the object on a fixed point on the hand, such as the thumb. A s explained in the experimental set up no objects could be manipulated without the addition of a gripping, high friction surface to the finger tips. The smooth surface created on the finger tips during the 3D printing process is poorly suited for any kin d of dexterous or gross mani pulations. The Raptor surpassed the Raptor Reloaded in its ability to complete the most tasks, with a 60% completion rate, versus a task completion Raptor Reloaded. The wider fingertips and palm allowed for more surface area to be applied to the objects for more secure manipulation. The decreased sized of the revised Raptor Reloaded is a detriment to its overall ability to perform grasping tasks, a nd wa s a strength during the lateral failure mode tests indicated that an improvement was not made across the hands design iteration. The Raptor Reloaded did not perf orm as well as the Raptor in either test. The Flexy Hand consistently failed tasks that require d the manipulation of the larger objects of the SHAP tasks. The fixed thumb in a f orward adducted position reduced the overall size the hand can be opened comp ared to the Raptor and Raptor Reloaded, which are designed to be in a naturally abducted posture.

PAGE 82

72 The failures experienced during the completion of the SHAP tasks shows that the hands do not have the long term durability that is an essential requiremen t for prosthetics. While these failures may have been a nuisance that could be quickly remedied in a workshop setting, they woul d be a major issue if a user we re wearing them day to day. The small size of the flexion line requires precise tweezers and gr asping tools to effectively tie knots. The complex knots require d experience to tie p roperly, and as demonstrated, did not guarantee a secure fixation. Re anchoring tension pins into th eir corresponding screw required a screwdriver, a tool that may not b e readily available. This issue was encountered again when grip strength testing was attempted Due to these issues, no reliable or consistent grip strength data could be gathered, as the hands were unable to apply a consistent load without an error aris ing that would invalidate the test. i.e., knot slippage, tension pin pullout. In a real world scenario, the wearer would have to forgo functionality of their device until repairs could either be made at home by the family, or if they were incapable of mak ing the repairs themselves, by their specialist, an even bigger inconvenience that would likely lead to device abandonment. Conclusion The results of the tensile testing showed that the ABS samples were superior material to the photopolymer ABS substitut e RGD525. While RGD525 was stronger, its brittleness leaves it ill suited for prosthetics, where robustness and ability to withstand everyday wear and tear is a necessity. In this study, ABS and RGD525 w ere compared to 3D printed composites as well. Whi le the performance of 3D printed carbon fiber was significantly greater than the examined, their costs is still a major hurdle to those seeking a low cost alternative to traditional prosthetics. Two material brands were compared, and

PAGE 83

73 printing parameters w ere controlled during all tests, but there are other parameters of the materials and 3D printing process that were not. The color of the filaments used in the ABS trials was not controlled for. The ambient environment in which the FDM prints were made wa s not controlled, i.e., ambient temperature, humidity, barometric pressure. Further studies could examine these effects and how they may apply to producing 3D printed prosthetics. Only two print orientations were examined during tensile testing, should be examined in future studies. Results of the lateral failure mode for a standardized test method of hands show ed that strain rate wa s not a major consideration for the 3D printed components of prosthetic devices made in ABS. The consistency and accessibility of the results gathered c ould be readily translated into other modes of failure, such as distant finger joints, wrist joints, or other orientations of the proximal joint. The simplicity of the test set up makes it easily accessible to others looking to evaluate their own 3D printed hand designs, or test other designs not examined in this study. The next step in creating a standard would be to partner with either standards des ignating body, such as ASTM, or a recognized leader in prosthetics research, in either academia or industry, to do further tests on other devices designs that were excluded from this study. The results of this test from a wide arra y of hands w ould allow f or a consensus to be reached on what values are acceptable for robustness and durability that is suitable for use in daily life of patients. Hands compared against the standard could be given a pass/fail, allowing for redesigns or rejection. The lack of coordination between independent device designers and medical professionals can be solved if standards that address the concerns of both groups become

PAGE 84

74 available. This study lays the groundwork for such a standard to become widespread, and offers a method of testing that can be built upon and remain accessible. Only 4 e NABLE hand designs were examined in this study, while dozens more are available open source through various sources. This standard testing method c ould be applied in further studies, wher e a database of results c ould be constructed, giving a comparison of the robustness of all available designs. This could be a source of reliable and verifiable knowledge for clinicians and specialist to decide which hand designs are best suited for their patients. Assessment of th e SHAP tasks with the three selected hand designs showed that current 3D printed hand designs have a very limited range of motion and postural grip patterns they can achieve, or at least mimic. The failures that occurred to each design during the tasks is even more telling, showing that these designs are not ready for long term use. While inexpensive and colorful, the hands are awkward to use, cannot support heavy objects, and have many points of failure. In their current state, the designs should be con sidered little more than toys, and their functionality compares poorly to traditional prosthetic devices. However, t hey are colorful, inexpensive, and aesthetically pleasing, and could serve as an important aid in helping a limb deficient child develop so cially and prepare them for fitting and use of a traditional prosthetic later in life. This in itself is an important goal for prosthetics and specialists in treating pediatric amputations, but further work must be done to improve the reliability and robu stness of 3D printed hand designs, if an inexpensive and functional prosthetic for children is to become a reality.

PAGE 85

75 R EFERENCES 3D Printable Prosthetic Devices (2017, 10 5). Retrieved from NIH 3D Print Exchange: https://3dprint.nih.gov/coll ections/prosthetics Acrylonitrile Butadiene Styrene (ABS) Typical Properties Generic ABS (2017, August 18). Retrieved from UL Prospector: https://plastics.ulprospector.com/generics/1/c/t/acrylonitrile butadiene styrene abs properties processing ASTM Inter national. (2014). Standard Test Method for Tensile Properties of Plastics. West Conshohocken: ASTM International. Beginner's Guide to 3D Printing (2017, August 18). Retrieved from 3D Insider: http://3dinsider.com/3d printing guide/ Biddiss, E. A., & Chau, T. T. (2007). Upper limb prosthesis use and abandonment: A survey of the last 25 years. Prosthetics and Orthotics International 236 257. Brusilovsky, D., Gothait, H., Levy, A., & Napadensky, E. (2004). United States Patent No. WO 2004096514 A3. Davids, J. R., Wagner, L. V., Meyer, L. C., & Blackhurst, D. W. (2006). Prosthetic Management of Children with Unilateral Congenital Below Elbow Deficiency. Journal of Bone and Join Surgery, 88 (6), 1294 1300. e NABLE. (2017, August 17). About us Retrieved from En abling the Future: http://enablingthefuture.org/about/ EOS. (2017, August 17). Systems and Equipment for Metal Manufacturing Retrieved from EOS e Manufacturing Solutions: https://www.eos.info/systems_solutions/metal Gothait, H. (1999). United States Patent No. 6,259,962 B1. I heart RC Hobby. (2017, April 28). Retrieved from I heart RC Hobby: http://iheartrchobby.com/ ISO 10328:2016 (2017, August 17). Retrieved from International Organization for Standardization: https://www.iso.org/st andard/70205.html ISO 22523:2006 (2017, August 17). Retrieved from International Organization for Standardization: https://www.iso.org/standard/37546.html Kate, J. t., Smit, G., & Breedveld, P. (2017). 3D Printed Upper Limb Prosthesis: A Review. Disabilit y and Rehabilitation: Assistive Technology 300 314. Krausz, N. E., Rorrer, R. A., & Weir, R. F. (2015). Design and Fabrication of a Six Degree of Freedom Open Source Hand. IEEE Transactions on Neural Systems and Rehabilitation Engineering

PAGE 86

76 Krebs, D. E., E delstein, J. E., & Thornby, M. A. (1991). Prosthetic Management if Children with Limb Deficiencies. Physical Therapy, 71 (12), 920 934. MarkedForge Materials (2017, August 18). Retrieved from MarkedForge: https://markforged.com/materials/ Metcalf, C. D., W oodward, H., Wright, V., Chappell, P. H., Burridge, J. H., & Yule, V. T. (2008). Changes in Hand Function with Age and Normative. The British Journal of Hand Therapy, 13 (3), 79 83. Murphy, K. (2017, February 17). College students build 3D printed prostheti c hands for kids. USA Today NinjaTek. (2017, September 11). Retrieved from NinjaTek: https://ninjatek.com/ Polylactic Acid (PLA) Typical Properties (2017, August 18). Retrieved from UL Prospector: https://plastics.ulprospector.com/generics/34/c/t/polylac tic acid pla properties processing ProductGoGo. (2017, April 28). Retrieved from ProductGoGo: http://www.productgogo.com/ Resnik, L. (2011). Development and testing of new upper limb prosthetic devices: Research designs for usability testing. Journal of Rehabilitation Research and Development 697 706. Shetty, P., Menezes, L. T., Tauro, L. F., & Diddigi, K. A. (2013). Amniotic Band Syndrome. Indian Journal of Surgery 401 402. Smith, D. G. (2006, January). Limb Loss in Children: Congenital Limb Deficienci es & Acquired Amputations. inMotion, 18 (1). Stratasys. (2017, August 31). Compare PolyJet Materials Retrieved from Stratasys.com: http://www.stratasys.com/materials/polyjet/compare polyjet materials Vasluian, E., Bongers, R. M., Reinders Messelink, H. A., Dijkstra, P. U., & van der Sluis, C. K. (2014). Preliminary study of the Southampton Hand Assessment Procedure for Children and its Reliability. BMC Musculoskeletal Disorders 199 212. Ventola, C. L. (2014). Medical Applications for 3D Printing: Current a nd Projeted Uses. Pharmacy and Theraputics 704 711. Zuniga, J., Katsavelis, D., Peck, J., Stollberg, J., Petrykowski, M., Carson, A., & Fernandez, C. (2015). Cyborg Beast: A Low cost 3D printed Prosthetic Hand for Children with Upper Limb Differences. BMC Research Notes 8 10.