‘'‘‘Z Department ofVeterans Affairs

Journal of Rehabilitation Researchand Development Vol . 31 No . 1, 1994Pages 42—49

Automated fabrication of mobility aids : Review of the AFMAprocess and VA/Seattle ShapeMaker software design

David A . Boone, CP; J. S . Harlan, BSME; Ernest M. Burgess, MDProsthetics Research Study, Seattle, WA 98122

Abstract—Computer-aided design and manufacture ofprosthetic and orthotic devices has recently moved out ofthe laboratory into clinical use . The Department ofVeterans Affairs Rehabilitation Research and Develop-ment Service has directed coordinated evaluation anddevelopment projects of this emerging technology underthe name: Automated Fabrication of Mobility Aids orAFMA . One of the major results of the effort was thecreation of mobility aid design software (ShapeMakerTM)concurrent with AFMA clinical testing of preexistingsystems and development efforts . In order to provide afoundation for future discussions regarding AFMA, thispaper provides a descriptive review of the AFMAprocesses and a review of the conceptual basis for theShapeMaker AFMA software development.

Key words : AFMA, computer-aided design, computer-aided manufacturing, computer aided socket design,orthotics, prosthetics.

INTRODUCTION

Computer software and computer controlledmachinery have been developed expressly to facili-tate the design and fabrication of prostheses andorthoses . Computer-Aided Design and Computer-Aided Manufacturing (CAD/CAM) (1) are parts ofthis new clinical development known as the Auto-

Address all correspondence and requests for reprints to : David A.Boone, CP; Prosthetics Research Study ; 720 Broadway; Seattle, WA98122.

This work was supported by the Department of Veterans AffairsRehabilitation Research and Development Service, Washington, DC .

mated Fabrication of Mobility Aids, or AFMA . ,The term AFMA was coined in January of 1987 byErnest M. Burgess, MD at a joint VA/NASAresearch meeting convened at the Langley NASAresearch center . AFMA has been our preferred termfor all related projects supported by the VA, sinceCAD/CAM is closely associated with an engineeringcontext which does not reflect the breadth anduniqueness of this clinical application . AFMA tech-nology is intended to improve service to the disabledin need of a mobility aid by increasing productionefficiency of the prosthetist/orthotist, thereby reduc-ing the time and financial burdens of fabricatingcustom prosthetic and orthotic devices . AFMA alsoprovides prosthetists and orthotists with a new toolthat has the benefit of numerical accuracy, consis-tency, and reproducibility that are essential toadvancing the science of mobility aid design.

The first comprehensive effort to develop aCAD/CAM prosthetic design system was begun byMr. Jim Foort, Director of the Medical EngineeringResource Unit (MERU) of the University of BritishColumbia . Foort is considered the progenitor of theentire concept (2) . The MERU group, in collabora-tion with the Bioengineering Centre of UniversityCollege London (UCL) in England demonstrated acomplete working AFMA system at the WorldCongress of the International Society for Prostheticsand Orthotics in London in 1983 . At the 1986 ISPOWorld Congress in Copenhagen, the UCL groupdemonstrated a newly developed Computer AidedSocket Design (CASD) software and hardware sys-

' Prosthetics Research Study Report listed in the Rehabilitation R&DProgress Reports-1987.

42

43

BOONE et at .

AFMA : Review of Process and Software Designs

tern that would be later tested in the VA Rehabilita-tion Research and Development Service (RehabR&D) National AFMA Research Project (3) . De-spite limitations, this generation of design softwareshowed promise in VA sponsored clinical trials (4)(Figure 1).

In 1985, the US Veterans Administration (VA),sponsored an international AFMA workshop whichbrought together the leading AFMA researchersfrom Canada, England, and the United States,along with interested investigators working with theVA Rehab R&D Service . Following the meeting, VARehab R&D accelerated support for a wide range ofprojects pertaining to AFMA. The centrally directedresearch culminated in a national cooperative effortamong Prosthetics Research Study (PRS) in Seattle,the Prosthetics Research Laboratory of Northwest-ern University in Chicago, and Rehabilitation Engi-neering Research Program of the New York VAMedical Center, (NYVAMC) . One primary objectiveof the National AFMA Research Program wasclinical use and evaluation of the UCL CASD, and,to a limited degree, the MERU CANFIT systems.This testing program resulted in the development ofa stand-alone prosthetic cast digitizer and designsoftware dubbed ShapeMaker (5).

This is the brief lineage of VA AFMA softwareand hardware development and is not intended to bea complete historical review of the field . Rather, wepresent a discussion of the conceptual and practicalconsiderations that are the foundation upon whichAFMA and the VA/ShapeMaker software is based .

METHODS

It will be helpful to the reader to review theAFMA process using the example of prosthetics as abasis for later discussion of AFMA software design.The basic principles may be extended to the designand fabrication of a wide range of mobility aidssuch as custom seating and foot orthotics.

In the design stage of the AFMA process thereare three main tasks : input of the anatomical form,design of the prosthesis, and output of the finisheddesign to a fabrication system (Figure 2).

First, the residual limb geometry is entered intothe computer so that an accurate graphical represen-tation of the limb is displayed on the computerscreen. This process is called digitizing . Typically,digitizing begins by taking a plaster negative cast ofthe residual limb using standard plaster castingbandage (Figure 3) . The cast of the limb is thenplaced in a device (digitizer) that digitally measures aseries of horizontal profiles on the inner surface ofthe cast as well as anatomical landmarks labeled bythe prosthetist (Figure 4) . A computer in communi-cation with the digitizer translates the numericalmeasurements into a three-dimensional representa-tion of the limb. Other technologies for digitizingthe limb shape directly are available, but to date themethod of digitizing casts has been incorporatedinto AFMA because of lower cost and applicabilityto digitizing many levels of lower and upper extrem-ity residual limbs.

After digitizing, the second step in the processis to sculpt the residual limb geometry seen on the

Figure 1.Schematic screen representation of prosthetic socket designparameters in UCL CASD software .

Figure 2.Diagram of the three primary steps in the AFMA process .

44

Journal of Rehabilitation Research and Development Vol . 31 No. 1 1994

Figure 3.Plaster wrap cast being made of a below-knee residual limb.

Figure 4.Plaster wrap cast in electromechanical digitizer designed byPRS.

computer screen into an acceptable prostheticsocket . Modifications to the digitized limb shape arenecessary to design a socket capable of comfortablysupporting the amputee . Following traditional prac-tice, specific alterations to contours and volumes ofthe digitized limb are made to create a comfortablesocket by supporting the body with compressingforces in areas of the residual limb which can

tolerate this stress and relief over those areas whichare stress intolerant (Figure 5) . In both cases, theshape modification is in the form of a smoothdeformation contour, either a depression or a relief.This should not suggest that we believe that thedeformations based on traditional clinical practicecreate ideal tissue loading characteristics . Lackingknowledge concerning optimal tissue loading pat-terns, the software mimics current practice as amatter of expediency . Using AFMA systems tocontrol for design variables may be an importanttool for further research in optimizing stress distri-butions in mobility aid design.

The third computer-based task is to translatethe completed socket design shown on the computerscreen into a "real object," in this case theprosthetic socket . This manufacturing process isaccomplished through computer-controlled carvingof a solid model of the socket design shown on thecomputer screen (Figure 6) . Once this model iscomplete, a thermoplastic prosthetic socket isformed over the positive using a standard vacuumforming process (Figure 7) . After the plastic hascooled, the solid model is removed leaving thenegative plastic impression for the socket . Theformed plastic socket is attached to the rest of themodular endoskeletal structure of the prosthesis(Figure 8) . The amputee dons the prosthesis andwith the addition of a suspension aid is ready forambulation.

Figure 5.Modifications to contour to provide supporting or relievingforces within a prosthetic socket are coded as colors on thecomputer screen .

45

BOONE et at .

AFMA: Review of Process and Software Designs

Figure 6.A three-dimensional model being carved from a plaster and cornstarch mixture that duplicates the form on the computer screen.

Figure 7.Thermoplastic socket being formed over the carved model in anautomated process oven.

The clinician uses AFMA software such asShapeMaker to accomplish all three of the tasksmentioned above : digitizing the residual limb shape,rapid design of a custom mobility aid based onclinical assessment and judgment, and, finally,computer-controlled fabrication .

Figure 8.The modular DVA/Seattle Limb System with AFMA socket.

DISCUSSION

The VA National AFMA Research Programresulted in the validation of the AFMA methods andidentification of areas needing further research anddevelopment (4,5) . For example, Houston, et al . hasshown that the method of using standard patterns ofmodifications superimposed on an individual digi-tized residual limb form coupled with prosthetistmanipulation of the shape using only the computerwas more effective in creating prosthetic socketsthat, in the amputee subjects' estimation were,"better fitting" and "more comfortable" thanmanually manipulated socket designs (5) . Thesestandardized modifications within the UCL-CASDsystem being tested were called rectification pat-terns. They contained regions in which the contourof a limb shape digitized in a cylindrical coordinatesystem were deformed inward or outward by mathe-matical addition or subtraction of radius values atcoordinates and in amounts predetermined in thedefinition of the rectification pattern. These alteredregions corresponded to areas where the prosthetisttraditionally modifies the plaster positive model ofthe socket using manual sculpting tools and wetplaster . However, while the magnitude of modifica-tion over each region could be changed in the CASD

46

Journal of Rehabilitation Research and Development Vol . 31 No . 1 1994

system, position, orientation, and contour of theregion could not be changed by the user . Addition-ally, the definition of new rectification patterns wasdifficult with this system (6) . Based on this andother observations of the National AFMA ResearchProgram, a "wish-list" pertaining to software wascreated, thus providing the basis for the AFMAsoftware development project initiated by Prosthet-ics Research Study in March of 1989 . 2 Some of themany software requirements outlined were: 1) thesoftware should run on a standard computer plat-form without requiring additional hardware orsoftware; 2) the application should be designed to becapable of creating most mobility aid forms ; 3) thesoftware should be as unobtrusive as possible inachieving clinical results ; 4) the prosthetist/orthotistshould be able to modify the surface as a realisticthree-dimensional projection rather than as a cross-sectional or schematic representation ; 5) the usershould have non-modal access to surface manipula-tion tools with varying degrees of fineness ; and, 6)the software should have an easy method forautomating replication of the technique of anindividual user . Software intended to meet theseobjectives has been completed by PRS and iscurrently in clinical use under the name ofShapeMaker . Following is a discussion of theAFMA software design considerations which forgedShapeMaker into its present form.

User InterfaceOne of the most important goals in developing

ShapeMaker was to limit the amount of newlearning that was required of the clinician . It wasour intent that aside from the mechanics of interact-ing with the computer using a mouse cursor control-ler, the prosthetist should not have to become acomputer expert to utilize the computer for pros-thetics . This is what is commonly called "user-friendly ." Our basic premise with regard to userinterface considerations was that the patient'sprosthetist/orthotist already knew what to do clini-cally for him or her and that the software should aidrather than interfere with that process.

The goal of developing easy to use softwarealso governed the choice of a computing platform.Our investigation considered the options of using ahigh-end UNIX-based graphics workstation, a DOS-

based PC, and the Macintosh II series. TheMacintosh II was selected for its wide availability,consistent and "user-friendly" interface, and excel-lent graphics performance without additional hard-ware or software . The choice of the Macintosh 3 inturn had a profound effect on the design of thesoftware by providing a consistent user-interface (7).

Input and OutputThe first and third tasks of the AFMA process

described above pertain to the input of the limbshape into the computer and output of the modifiedshape for fabrication . ShapeMaker has been pro-grammed to be "machine and resolution indepen-dent" and is able to communicate with mostcommercially available digitizing and carving devicesthrough serial and parallel interfaces . In addition,ShapeMaker can also communicate with equipmentlocated at a remote site via modem connection overtelephone lines . To extend the utility of the soft-ware, newly developed digitizing and carving hard-ware is similarly interfaced directly to theShapeMaker program through the addition of ma-chine specific modular software drivers . The task ofmobility aid design accounts for most of theprogramming that comprise ShapeMaker, and therest of this discussion is devoted to this task of theprocess.

Generic ModificationOne goal in AFMA software design has been to

create a software paradigm in which clinicians relateto the graphical representation of prosthetic andorthotic shapes in the same way they relate to thehand-sculpted solid plaster models where material iseither removed or built up, using tools to modify theoriginal limb shape . This is one way in which to easethe transfer of clinical skills from the traditionalfabrication system to a computer-based one.

ShapeMaker acts as a generic tool chest allow-ing the user to sculpt general three-dimensionalanatomical forms. Since it is not simply an above-knee prosthetics program or an orthopedic shoeprogram, ShapeMaker helps to fulfill the conceptthat AFMA technology is applicable to nearly allmobility aids . As such, our laboratory has used theprogram for most levels of prosthetic sockets as well

3 A version of ShapeMaker has also been developed using the2 Report listed in the Rehabilitation R&D Progress Reports-1989.

Macintosh-like Windows TM graphical user interface .

47

BOONE et at,

AFMA: Review of Process and Software Designs

as the design of prosthetic cosmesis, knee braces,body jackets, wheelchair seating, shoe insoles, shoelasts, and maxillofacial modeling. ShapeMaker isappropriate for all of these different applicationsbecause each process requires the basic task ofthree-dimensional shape deformation for which thesoftware was designed . Also, because the samesoftware can be used for all of these differentapplications, the user interface is the same for each.

Levels of ControlAutomated creation of mobility aids for

broadly defined patient populations was necessaryto achieve the goals of efficiency and consistencyafforded through automation . However, in practicethere is variation between individual patient anat-omy that we feel is most efficiently accommodatedthrough the exercise of clinical judgment and skill toindividualize a patient's custom mobility aid design.And so, a dilemma of control versus efficiency arosein the design of ShapeMaker . The program isdesigned to allow the user to very quickly applybroad modifications to the mobility aid design;however, the user also needs to have tools availablefor very fine, detailed manipulation of the shape.With global controls, clinicians might feel toorestricted at times, while in other cases they mightfind detailed controls too time-consuming or cum-bersome for general use . This defines a trade-offbetween flexibility and structure . In order to satisfysuch divergent design requirements, ShapeMakerwas developed around the concept of making differ-ent "levels of control" available to the clinician atall times.

ShapeMaker provides three distinct levels ofcontrol which could be labeled Automation, Modifi-cation, and Creation . Each successive level ofcontrol requires more interaction with the softwaretool, yet yields finer control over the finishedproduct . A good analogy would be different gradesof sandpaper . Rougher grades are used for shaping,while finer grades are used for finishing details.ShapeMaker utilizes a non-modal design whichallows the user to use functions from any level ofcontrol at any time.

The most general controls generally involvecreating predefined modifications very quickly, whatwe term automation. The user does not need tothink about fine details, but rather about the "bigpicture." Automation completes a large part of the

work toward a finished result with only a very smalleffort on the part of the user, such as completesocket modification at the selection of a singlecomputer command. For example, the rectificationpatterns in the UCL CASD software and theanalogous Template commands in ShapeMaker pro-vide this broad level of control . The result of acommand at the Automation level of control mayencompass many different actions that could also beaccomplished individually using finer levels of con-trol, but that at this level are completed automati-cally. The complex quantitative modification can bedescribed qualitatively in just a few words, "Createa PTB socket for this particular Trans-Tibial limbshape" (Figure 9) . Because of individual variationamong patients, we anticipate that only a smallpercentage of mobility aids will be finished usingonly the Automation level of control . The softwaremay also be programmed to take individual patientcharacteristics into account at this level to achieve adegree of automated customization . For example,ShapeMaker automatically interprets the angularorientation of the long bones relative to the surfacetopology in order to orient modifications moreaccurately.

At a finer level of control, the user can takemore direct control of the finished product bymodifying the results created using the Automationtools. For instance, after applying a socket designtemplate for a below-knee prosthesis, the user might

Figure 9.The Automation level of control is used to produce all thechanges between the two shapes shown using the single softwarecommand : Apply Template .

48

Journal of Rehabilitation Research and Development Vol . 31 No. 1 1994

wish to increase the depth of the patellar tendonmodification or to change the volume, smoothness,or length of the prosthetic socket . Again, themodification can be readily and accurately describedin a few words, as in : "Give 2 mm more relief overthe head of the fibula" (Figure 10) . Surface alter-ations of this type are generally quantitative orspatial orientation adjustments to modificationsmade using Automation tools . This level of controlwill achieve further gains toward the desired endproduct with only a small increase in the effortrequired by the user . As demonstrated in an evalua-tion of the UCL CASD system, many prostheticsocket modifications can be finished using thesecond level of control (4).

A third level of control gives the user morecomplete creative control over the finished product.New modifications of the surface may be freelycreated using dozens of specialized commands inShapeMaker . The user may manipulate the shapewith detailed control limited only by the resolutionof the surface data, which in itself may be manipu-lated in software by resampling the surface at adifferent coordinate resolution. The most importantand powerful command at this level of control iscreation of templates that would easily and com-pletely replicate a clinician's own modification tech-nique in a way similar to the UCL CASD rectifica-tion patterns . One of the principle shortcomings ofprevious software developments was that each pro-gram included only a small number of automatedsocket design techniques from which to choose.

Figure 10.A ShapeMaker modification region highlighted on the three-di-mensional socket model .

Every time the prosthetist designed a socket on thecomputer, the program would modify the socketaccording to one of the limited techniques . Unfortu-nately, this scheme forced the prosthetist to adapt toa foreign technique that might be subtly or evenradically different from his own design style . Tosolve this problem, software that would learn fromthe clinician's own technique was required . Becausedifferent clinical situations require different tech-niques, the prosthetist or orthotist is able to create alibrary of personally created techniques, or Tem-plates, appropriate for each particular situation.These templates then provide the basis for theAutomation level of control.

Early experiences with software for mobility aiddesign often encountered vocal opposition frompractitioners who exclaimed, "No computer is evergoing to make a socket as good as one made by askilled hand!" Given the conceptualization of thesoftware tool as an extension of clinical decision-making, this sentiment was misplaced because thecomputer is not creating the socket . One may usethe analogy of the software as a tool and thecomputer as a workbench that the clinician uses tocreate a mobility aid . With all of its levels ofcontrol, ShapeMaker allows the clinician to auto-matically apply his/her own techniques to the designof the mobility aid and then gives the clinician theresponsibility and capability of fine-tuning thesocket shape to suit the individual patient . Andfinally, using the creative level of control, the users,can take as much time as they are willing to spend tocreate an optimal design and then store the design asa template for future use . The time required at thislevel is generally longer than use of simpler levels ofcontrol, and creation of surface modifications gen-erally requires more time than using tools at moregenerally levels . But as in the Modification level ofcontrol, the emphasis is still the translation of aclinical skill to the computer tool, and not thelearning of an entirely new skill.

CONCLUSIONS

From the outset, AFMA software developmenthas occurred in order to provide clinicians andresearchers with control over the variables inherentin mobility aid design . Design software necessarilyinteracts with a user, and as it is used and the user

49

BOONE et al .

AFMA: Review of Process and Software Designs

gains competency and insight in use of the softwaretools, the relationship of user to how software isused changes . We conclude that the CASD andCANFIT software which prompted development ofthe VA/ShapeMaker software were too limiting tothe experienced clinician and did not adequatelyaccount for the creative aspects of individualizingmobility aid design . Because of this, AFMA soft-ware should be constructed within a paradigm of themultiple levels of control which accounts for auto-mation as well as fine artistic manipulation.

Computer-aided design and fabrication, as it isdescribed in this paper, is probably only a beginningfor the use of computers in prosthetics andorthotics. This goal has been achieved to somedegree, but software development rarely has adistinct end-point . Current and future developmentspromise to assist clinicians with tasks other thandesign, such as static and dynamic alignment,integrated digital video, and tissue stress analysis.Clinician access to user-friendly computer-basedtools will give more information and control to therehabilitation team and will benefit the many peoplewho require mobility aids.

ACKNOWLEDGMENTS

The authors wish to thank all of the members of theVA National AFMA Research Group in Seattle, Chicago,

and New York for their input and encouragement duringthe development of DVA/Seattle ShapeMaker . We arealso indebted to the amputee research subjects and theprosthetists across the nation who through their participa-tion ensured that software development was always basedin the reality of clinical necessity.

REFERENCES

Klasson B . Computer aided design, computer aidedmanufacture and other computer aids in prosthetics andorthotics . Prosthet Orthot Int 1985:9:3-11.

2. Foort J . The Knud Jansen Lecture : Innovation in pros-thetics and orthotics . Pros Orthot Int 1986 :10:61-71.

3. Dewar M, Jarman P, Reynolds D, Jones K . Clinical trialof the UCL computer aided socket design system.Bioengineering Centre Report 1986, University CollegeLondon, 1986 :13-16.

4. Boone D, Burgess E . Automated fabrication of mobilityaids : clinical demonstration of the UCL computer aidedsocket design system . J Prosthet Orthot 1989 :1(3) :187-90.

5. Houston, V, Burgess, E, Childress D et al . Nationalprogram for the automated fabrication of mobility aids(AFMA) : below-knee CASD/CAD testing and evaluationprogram results . J Rehabil Res Dev 1992:29(4) :78-124.

6. Dewar M, Reynolds D . Development of The UCLcomputer aided socket design system . BioengineeringCentre Report 1986, University College London, 1986 :11-12.

7. Human interface guidelines: the Apple desktop inter-face . Reading, MA : Addison Wesley Publishing Co . Inc .,1986 .