A Quantitative Method for Comparing and Evaluating ...

IEEE TRANSACTTONS ON REHABILITATION ENGINEERING, VOL. 4, NO. 4, DECEMBER 1996

303

A Quantitative Method for Comparing and Evaluating Manual Prosthetic Socket Modifications Edward D. Lemaire and Frank Johnson

Abstract-Manually designed prosthetic sockets are difficult to evaluate since the hand-sculpting modification process does not retain the initial shape for comparison. A quantitative method for defining and comparing manual socket modifications was developed and integrated into the CADVIEW software package. The numerical comparison procedure consisted of a) digitizing premodification and post-modificationmodels of a prosthetic socket, b) aligning these two shapes to a common axis (calculate cross section centroids, determine line of best fit through the centroids, rotate and move the line of best fit to a common axis), and c) displaying the differences in shape both numerically and using a color-coded three dimensional image. Alignment technique testing showed a between-radial-point average error of 2.5 mm using automatic alignment and 1.4 mm after further manual alignment adjustment. The between socket difference values were used to outline individual modifications and save these outlines to disk. Modification outlines from a series of patients were averaged to determine a prosthetist’s general modification style. Averaged results from two prosthetists qualitatively supportedthe effectiveness of this procedure. This alignment and comparison system should help transfer hand-sculpting skills to prosthetic CAD/CAM systems, clinical research, and education for new clinicians.

I. INTRODUCTION

P

Various approaches have been implemented to document changes in limb volume [14]-[16], changes in shape [15]-[17], and display changes made on a CAD/CAM system [18]; however, a method for quantitatively defining manual socket modifications has not been reported. A system for defining and displaying manual socket modifications would have immediate benefit for education since the student and teacher could see the initial shape, final shape, and areas that were handsculpted. This feedback should help the student visualize what modifications are required to successfully fit a prosthetic client. Quantification of manual modifications should also help with the transition from “hand-sculpting’’ to CAD/CAM socket design. Quantitative information specific to a prosthetist’s personal modification style could be copied to a prosthetic CAD system and used as a template for future modifications. This should decrease both CAD training time and frustration for the prosthetist. From a research perspective, quantification of manual modifications should provide a mechanism to compare prosthetic techniques and procedures between regions and/or countries, types of residual limb shapes, and different design philosophies. It can also provide a means for documenting modification techniques used by experienced prosthetists. This paper describes an addition to the CADVIEW software package [19] which takes digitized data from two prosthetic shapes, calculates the differences between shapes, and determines manual modification areas.

ROSTHETISTS have traditionally combined clinical guidelines with artisan skills to create a prosthetic socket. While materials have changed over the years, a socket shape is still obtained by hand-sculpting a model of the client’s residual limb to provide relief for sensitive areas and to apply locomotor forces to load resistant areas [l], [2]. While this technique is successful, socket shapes are often not reproducible by the particular clinician and are usually quite variable between prosthetists [3]. It is also difficult for a prosthetist to explain what modifications were made on a positive model since the initial shape and individual modifications are lost during the sculpting process. With the emergence of CAD/CAM (Computer Aided Desigdcomputer Aided Manufacture) as a socket design system, a quantitative method for modifying a socket model is available [4]-[ 121. These systems are efficient and reliable, provided that the prosthetist has sufficient training. A common problem with CAD/CAM training is the transfer of manual “hand skills” to virtual computer skills [13].

A. Data Input

Manuscript received August 2, 1995; revised June 27, 1996. This project was supported in part by the Ontario Rehabilitation Technology Consortium and the Labatt’s Relay Research Fund. E. D. Lemaire is with the Institute for Rehabilitation Research and Development, The Rehabilitation Centre, Ottawa, Ontario, K l H 8M2 Canada. F. Johnson is with the Medical Engineering Industry Research Affiliation, University of Ottawa, Ottawa, KIN 6N5 Canada. Publisher Item Identifier S 1063-6528(96)08805-2.

CADVIEW accepts digitized data files in the Univ. of Texas [I31 generic data format, CANFIT-PLUS data format (VORUM Research Corp.), and IPOS System IT output files. The data are stored as polar coordinates and radial distances for a series of transverse cross sections along the longitudinal axis of the shape. A clinician may digitize or laser scan a model

11. METHODS The prosthetic socket analysis program was developed for Microsoft Windows 3.1. The Microsoft Windows 3.1 interface makes this program easy to use, provides a means for copying on-screen images to other graphics or word processing programs, and provides support for most printers and graphics adapters. A multiple document interface was used so that more than one comparison shape could be simultaneously viewed on-screen. This feature is beneficial for viewing both manual and CAD/CAM socket modifications from the same patient.

1063-6528/96$05.00 0 1996 IEEE

304

IEEE TRANSACTIONS ON REHABILITATION ENGINEERING, VOL. 4, NO 4, DECEMBER 1996

of the residual limb or socket using commercially available hardware and controlling software. The data resolution for these systems is one to two millimeters (depending on the system). While any system can be used, the remainder of this paper will use examples based on the CANFIT-PLUS cast digitizing method.

B. Manual Shape Alignment Two approaches have been used for digitizing the residual limb and modified socket shapes. The most reliable approach involved casting a client's residual limb in the traditional method and filling the cast with plaster. Care was taken to ensure that the mandril was positioned along the longitudinal midline of the distal 3/4 of the cast. This permitted digitization of the shape even when excessive knee flexion was used in casting, although knee flexion should be kept to a minimum during the casting session. Mandril deviation from the cast's longitudinal midline did not affect the comparison process since both pre-modification and post-modification shapes had the same deviation. After the plaster had hardened, a nail was drive into the plaster positive at right angles to the main reference point. For a transtibial shape, the main reference point could be the patellar tendon band (PTB) or the tibial tubercle. A nail was required since material is typically removed form the patellar tendon region during modification. By driving the nail deeper into the model as material is removed, the main reference point position is maintained. The main reference point must be accurately defined since its normal coordinates will define the angular position during shape alignment. The positive model's mandril was positioned in the cast digitizer and the positive model was digitized using the standard protocol. Following initial digitization, the positive model was handmodified by the prosthetist. The modified positive was positioned in the digitizer in the same manner as the unmodified shape. Since the mandril had not been moved, vertical positioning of the two shapes was the same. If the main reference point has not changed between digitizations, the two shapes will be in the same orientation. Common data for both sockets were calculated by interpolating new post-modification data points at the same angular and longitudinal positions as the pre-modification socket coordinates. A cubic spline interpolation was used to calculate the new data points. C. Automated Shape Alignment

While the positive model method had the potential to give a more accurate alignment, the post-modification shape may not be the same as the final socket shape. Often, a prosthetist will make minor modifications directly on the socket during the fitting process. To assess the modifications needed to make the final socket, a clinician must digitize the initial cast (base shape) and the final socket (comparison shape). Unfortunately, these two digitized shapes will not be aligned to the same axis. Since various modification and manufacturing stages occur between casting and fitting, consistent landmarks (i.e., landmarks in areas that are not changed) are typically lost.

Fig. 1. Initial socket position and cross-section positions for shape alignment tests.

Fig. 2. Manually changing comparison socket rotation.

To deal with the lack of reliable landmarks, a shape based alignment process was implemented. Using the initial data file as the base, the initial and final shapes were aligned to a common axis along the longitudinal midline of the socket. The alignment procedure involved calculating the centroid of each transverse slice and the line of best fit through these centroids. All data points were rotated in the sagittal and coronal planes, about axes passing through the vertical and horizontal midpoints of the shape (i.e., origin), so that the line of best fit ran along the common vertical axis. Since the transverse slices were now at an angle to the main reference frame, new data points were interpolated to keep all cross sections in the transverse plane (longitudinal cubic spline interpolation). This process was repeated five times. Five alignment iterations were required since the socket may be, initially, at an angle with the vertical. A digitizer always selects values at right angles to the vertical; therefore, the line of best fit through the centroids of these horizontal cross

LEMAIRE AND JOHNSON: QUANTITATIVEMETHOD FOR COMPARING PROSTHETIC SOCKET MODIFICATIONS

305

Fig. 3. CADVIEW compare socket display

sections would not correspond to the longitudinal midline of the socket. With each iteration, the socket is closer to vertical and the centroidal line of best fit is closer to the socket’s longitudinal axis (since the new data points are obtained from a shape that is more vertical). Five iterations were found to provide an adequate compromise between execution speed and alignment accuracy. If the two shapes were very close to the correct alignment, the changes between iterations would be minimal. Comparison (final) shape alignment followed the same procedure as with the premodification shape; however, interpolated data points for the comparison shape were at the same vertical and angular locations as the base shape. Any slices beyond the limits of the base shape were not included in the comparison. Following this stage, each socket was defined by a series of points with common locations but different magnitudes. While this method worked for many sockets, cases can occur where the optimal center of rotation is above or below the origin. Also, the comparison socket could be linearly translated. Manual alignment tools were added to graphically move the origin on-screen, rotate the comparison shape relative to the base shape, or move the comparison shape relative to the base shape. This fine tuning was helpful for minor alignment corrections since the majority of the alignment processes were performed by the automated procedure. D. Alignment Evaluation The alignment method was tested by digitizing the same shape twice-once in the standard vertical position and once at a five to ten degree angle to the vertical (Fig. 1). The two shapes were aligned using the automated procedure and the root mean square error (RMSE) between the two shapes

was calculated from the common data points. The mean and standard deviation for five trials were calculated and compared to the base CAD/CAM modification unit (1.O mm-the minimum units used in the CANFIT-PLUS modification software program). Through clinical use, a 1.0 mm value has shown to provide adequate resolution for defining socket modifications. Following this test, the orientation adjustment tools were used to try to improve the alignment of the five test cases. RMSE statistics were also calculated and averaged for the adjusted alignment trials. To investigate the effect of modifications on the alignment procedure, a series of modifications were applied to a transtibial residual limb shape using the CANFIT-PLUS program. To test whether both the original and modified sockets would align to the same axis, between socket differences for axis rotation and position were compared. Since the CAD shape alignment did not change, variations in the RMSE were due to the alignment process. The RMSE for aligned and unaligned trials were compared. E. Graphic Display

Aligned shapes were displayed on-screen in one of four ways: overlapped wireframe, color mapped solid, color mapped overlay, and overlapped cross section. The overlapped wireframe view involved rendering wireframe images of both sockets on the same Cartesian axes. This view allowed a prosthetist to see differences in length and to detect and correct alignment errors. The color mapped view involved creating a blue scale and a red scale on the graphics color palette. The original shape was rendered in shades of blue or red relative to the differences in radial length between socket data points. Areas where the socket has been built up were displayed in shades of blue while

306

IEEE TRANSACTIONS ON REHABILITATION ENGINEERING, VOL. 4, NO. 4,DECEMBER 1996

Fig. 4. Outlining and comparing socket modifications.

areas where material was removed were displayed in shades of red. The maximum color scale value could be changed by the user, thereby providing good color contrast during evaluation. Minimum values could also be set so that areas with modifications magnitudes less than the minimum would be rendered white. Minimum display values were also used to define modification boundaries. The color mapped view was the best for examining socket modification patterns since the modification areas were well defined, modification contours could be visually examined, and the image was uncluttered. The color mapped overlay view involved rendering the color mapped view and the comparison socket's wireframe view on the same axis (Fig. 2). While this view can be helpful for locating general areas of modification, the display becomes cluttered as the mesh density increases (Le., more data points used to define the shape). To examine the differences between sockets at each crosssectional slice, an overlapped cross-section view was included. By rendering the base and comparison cross sections on the same axis, a prosthetist can examine the shape of the modification and define alignment errors. This two-dimensional (2-D) view also corresponds well with the 2-D views used for modifying sockets in commercial CAD/CAM systems. The overlapped cross-section view should help the prosthetist visualize how their manual modifications should look onscreen during a CAD session. Since a multiple document interface was employed, the position of the active cross section could be highlighted on the three-dimensional (3-D) image to show users the location of the overlapped slices. The ability to open more than one 2-D or 3-D comparison window was useful for comparing modifications at different longitudinal locations on the socket and when comparing the modifications on different sockets.

In addition to 2-D and 3-D graphics, the user can inspect the differences between sockets at each data point. When the mouse passed over a data location, the radial difference between sockets was displayed in millimeters at the bottom of the program window. This option was valuable when evaluating the amount of manual modification performed on the residual limb shape (Fig. 3).

F. Outlining" Modifications Socket editing on most commercial CAD systems involves drawing an outlinehoundary of the modification area onscreen, adding apexlcontrol points within the outline, defining the amount of modification at the apex point, and defining the contour of the modification region. To aid in the transfer of manual modification skills to CAD/CAM, a modification outlining feature was added to CADVIEW which automatically determines areas of modification and superimposes the defined outlines on the 3-D comparison image (Fig. 4). To define the modification areas, all peak points on the comparison shape were defined. These points corresponded to data values that were greater than (maximum peak) or less than (minimum peak) all surrounding data points. Starting from each peak point, data along the peak's cross section were iteratively evaluated until the radial distance became less than the minimum display value (i.e., the first boundary point). The remaining boundary points were located by following a contour line between data points greater than and less than the minimum display value. The boundary points, side of the amputation, and modification name could be saved to disk as a group or individually. While the automated outlining method was good for discrete modifications, many manual socket modifications were blended together. To adapt to the lack of modification sepa-

-

LEMAIRE AND JOHNSON: QUANTITATIVE METHOD FOR COMPARING PROSTHETIC SOCKET MODIFICATIONS

ration, various outline editing tools were added to allow the clinician to break a large modification into smaller areas and edit the shape of a modification. The “break modification” feature was useful for defining two blended areas that were initially treated as a single modification; for example, a fibular head modification could be blended into the posterior shelf or the anterior lateral prominence of the tibia. Another method for defining a blended modification was to increase the minimum display value until the modification area became independent, generate the outline using the automated procedure, reset the minimum display value to show the correct boundary locations, and use the outline editing tool to move the data point out to the correct locations.

307

TABLE I ROOTMEANSQUARE ERRORDUE TO PROSTHETIC FOR PREALIGNED CAD SHAPES SOCKET MODIFICATIONS ( N O AUTOMATIC ALIGNMENT AND AUTOMATIC ALIGNMENT)

TABLE I1 DIFFERENCES IN AXISALIGNMENT PARAMETERS WHEN COMPARING A NON-MOD AND MODIFIED SOCKET (ROTATION IN DEGREES, TRANSLATION IN MILLIMETERS)

G. Averaging ModiJications

Examining individual cases was valuable for defining the general modifications used by a prosthetist; however, individual differences between clients may not give the best representation of a prosthetist’s overall modification style. Most commercial CAD programs will allow the user to save a template/overlay of modifications and apply the overlay to other sockets in one operation. The ability to define a prosthetist’s general modification style should help with the creation of a template which best resembles hisher manual technique. After individually saving modification areas, common areas from several sockets were reloaded from disk for averaging. The data points on each outline were defined in millimeters (longitudinal location of the cross section) and radians (angular location on the cross section). Absolute values were required to take the size of the outline into account during averaging. The centroid for each outline was calculated and the centroid coordinates were subtracted from all outline data. All outline coordinates were now located about the origin of a Cartesian coordinate system. Common data points for each outline were interpolated at the intersection of a line from the origin to the outline. Average values were obtained from the common data points at 10 degree intervals about the origin. The new outline position was obtained by averaging the centroid coordinates from each individual outline and translating the new data points to the average centroid coordinate. The average outline was saved to disk and displayed on the 3-D comparison image. This outline could be redrawn on a commercial CAD program and saved as a template or, with the appropriate file format, saved directly as a template file.

H. Qualitative System Evaluation Using the socket comparison and modification outlining features, the modification techniques for two experienced prosthetists were analyzed. Seventeen transtibial sockets from Prosthetist A and eight sockets from Prosthetist B were used in the evaluation. Positive models from all sockets were digitized before manual modification and after manual modification. Using the comparison software, common outline areas from the 17 Prosthetist A shapes were averaged to create a representative modification pattern. This average modification pattern was used to produce an overlay on the

CANFIT-PLUS system by redrawing the averaged outlines and saving the outlines as a CANFIT-PLUS overlay. After applying the overlay to a residual cast shape, the modifications were visually assessed using CADVIEW. A color mapped view of the CAD modification was displayed beside a color mapped view of manual modifications performed on the same socket. The magnitude of the CAD modifications was obtained using the mouse/data inspection feature. The comparison features were evaluated based on alignment exactness, ability to see modification areas, ability to differentiate differences in technique, and ease of use. 111. RESULTS/DISCUSSION

The shape alignment test produced an average RMSE of 2.5 mm with a standard deviation of 1.3 mm. After performing minor changes to the comparison shape orientation, such as rotation in the sagittal and coronal planes, the intershape error was reduced to an average of 1.4 mm (standard deviation = 0.6 mm). These results were acceptable considering that the accuracy of the digitizer was 1 mm. Based on these results, modifications below 1 mm were not considered in the evaluation of socket modifications. The modified shape comparison results are listed in Tables I and 11. The effect of single and full shape modifications on the RMSE was less than 0.4 mm. This result is based on a difference in angular axis rotation of less than 1.5 degrees and, in all cases except the zy plane translation, translation differences of less than 0.4 mm. For the fully modified socket, the alignment error was reduced to 0.05 mm by manually translating the shape 2.0 mm in the z y plane. These results support the automatic alignment method for use with prosthetic sockets. A possible explanation for these results is that prosthetic modifications are made over the entire shape and not one location, allowing the line of best fit through the slice centroids to maintain approximately the

308

IEEE TRANSACTIONS ON REHABILITATION ENGINEERING, VOL. 4, NO. 4, DECEMBER 1996

Fig. 5. Comparison views from two prosthetists.

‘

same relative position. Since the volume of the residual limb is constant, socket modifications in one area must be countered by other modifications to ensure that the socket is not too tight or too loose. While the shape of individual cross sections will change, the net effect of these changes on the line of best fit is minimized (since changes in slice centroids position at one location are countered by opposite changes at another location). If a prosthetist typically makes very large modifications to one side of a socket, more manual alignment may be required to position the final shape. Qualitative feedback from the two prosthetists was positive. The software was very easy to use since people with some MS Windows 3.1 experience learned to compare two sockets onscreen within 15 min. Modifications to shape alignment were required for only one shape and were attributed to digitizer error. The shape alignment method and tools were considered easy to use and effective. The socket comparison feature was considered a useful tool for examining modification technique. Prosthetist A, with 26 years experience, had a very consistent modification technique in all areas except the patellar surface. One interesting result was that the prosthetist believed that he removed material equally along the posterior surface of the positive model. The comparison view, though, showed that he consistently removed more plaster on the right-hand side of the model (for both left and right leg amputees). This is probably due to the prosthetist gripping tighter with his right hand during the casting stage and correcting the shape during the modification stage. If this prosthetist made a standard CAD overlay for the posterior socket region, with the apex point in the middle of a large boundary, too much material would be taken off the

left side of the modification and not enough material would be taken off the right side. Differences between Prosthetist A and Prosthetist B were clearly visible (Fig. 5). These differences included positioning of the fibular head modification, blending the fibular head modification into surrounding buildups, size of the distal tibial crest modification, magnitude of the tibial crest modification, and modification of the posterior socket region. The initial overlay created from the averaged outlines was not satisfactory since the boundary areas were too small. Since the CANFIT-PLUS program requires larger boundary areas to facilitate blending of the modification into the socket, the overlay modifications were expanded and the comparison was repeated. Based on visual, on-screen assessment by the prosthetists, the final overlay successfully reproduced the manual modifications for the test socket. While this method provided a means for identifying individual and averaged socket modification patterns, the appropriateness of a single averaged pattern for all amputees is questionable. A more likely scenario is that a series of limb-type specific modification approaches will be identified. This software tool will be beneficial for discovering what these limb-type groupings are and for generating the series of average modification patterns specific to an individual prosthetist’s style. It is also unlikely that a prosthetic socket shape, generated from an averaged overlay, will be appropriate for all patients without some customization. However, the overlay method should get the prosthetists much closer to the final socket shape with one program operation. This process will reduce modification time and possibly increase product consistency

LEMAIRE AND JOHNSON: QUANTITATIVE METHOD FOR COMPARING PROSTHETIC SOCKET MODIFICATIONS

since the same starting point is used for each modification. Also, many CAD users find it easier to modify an existing shape than create the entire modification scheme from scratch.

IV. CONCLUSION A quantitative method was developed to define and compare manual prosthetic socket modifications. Qualitative and quantitative evaluation demonstrated that this method aligned socket shapes to an acceptable level of accuracy, displayed the modification information in an easily understandable and useable manner, and was able to produce an average modification pattern specific to an individual prosthetist. Quantitative socket comparisons should be useful for clinical evaluation, education of new prosthetists, research into clinical techniques, and transferring manual modification skills to CAD/CAM systems. ACKNOWLEDGMENT The authors would like to thank P. Bexiga and A. Platzer for clinical assistance and G. Martel for administrative assistance.

309

[lo] G. K. Ruder, “CAD/CAM transtibial temporary prosthesis: Analysis and comparison with an established technique,” Prosthet. Orthot. Int., vol. 16, pp. 189-195, 1992. [I11 C. G. Saunders and M. Bannon, “Results of external clinical trials of the MERU “CANFIT computer-aided socket design system,” in Proc. ISPO Sixth World Congress, Kobe, Japan, 1989, pp. 112. [I21 N. E. Walsh, J. L. Lancaster, V. W. Faulkner, and W. E. Rogers, “A computerized system to manufacture prostheses for amputees in developing countries,” J. Prosthet. Orthot., vol. 1, pp. 165-181, 1989. [ 131 A. K. Topper and G. R. Femie, “An evaluation of computer aided design of below-knee prosthetic sockets,” Prosthet. Orthot. Int.. vol. 14, no. 3. pp. 136142, i990. 1141 I. H. Bednarczvk, C. Hershler, and D. G. Cooper, “Development and clinical evaluation of a computerized limb volume measurement system (CLEMS),” Arch. Phys. Med. Rehab., vol. 73, pp. 60-63, 1992. [15] R. J. Belsole, D. R. Hilbelink, J. A. Llewellyn, S. Stenzler, T. L. Greene, and M. Dale, “Mathematical analysis of computer carpal models,” J. Orthop. Res., vol. 6, no. 1, pp. 116-122, 1988. [I61 G. R. Fernie, A. P. Halsall, and K. Ruder, “Shape sensing as an educational aid for student prosthetists,” Prosthet. Orthot. Znf., vol. 8, no. 2, pp. 87-90, 1984. [17] R. E. Borchers, D. A. Boone, A. W. Joseph, D. G. Smith, and G. E. Reiber, “Numerical comparison of 3-D shapes: P6tential for application to the insensate foot,” J. Prosthet. Orthot., vol. 7 , no. 1, pp. 29-34, 1995. [ 181 J. A. Sidles, D. A. Boone, J. S. Harlan, and E. M. Burgess, “Rectification maps: A new method for describing residual limb and socket shapes,” J. Prosthet. Orthot., vol. 1, pp. 149-153, 1989. [19] E. D. Lemaire, “A CAD analysis programme for prosthetics and orthotics,” Prosthet. Orthot. Int., vol. 18, pp. 112-117, 1994. ~~

REFERENCES 111 B. J. Alcock and R. G. Redhead. “Interface problems and possible solutions,” Prosthetics and Orthotics Practice. London: Edward Arnold Ltd., G. Murdoch, Ed., pp. 213-218, 1970. M. J. Quigley, Atlas of Limb Prosthetics: Surgical, Prosthetics, and Rehabilitation Principles. St. Louis, MO: Mosby-Year, J. H. Bowker and J. W. Michael, Eds., pp. 67-79, 1992. B. Klasson, “Computer aided design, computer aided manufacture and other computer aids in prosthetics and orthotics,” Prosthet. Orthot. Int., vol. 9, no. 1, pp. 3-11, 1985. J. R. Engsberg, G. S. Clynch, A. G. Lee, J. S. Allan, and J. A. Harder, “A CAD/CAM method for custom below-knee sockets,” Prosthet. Orthot. Int, vol. 16, pp. 183-188, 1992. V. W. Faulkner, N. Walsh, and N. Gall, “CAD/CAM of lower extremity prostheses: The San Antonio system, J. Rehab. Res. Dev., Progress Report, p. 6, 1986. V. L. Houston, E. M. Burgess, D. S. Childress, H. R. Lebneis, C. P. Mason, M. A. Garbarini, K. P. LaBlanc, D. A. Boone, R. B. Chan, 3. H. Harlan, and M. D. Brncick, “Automated fabrication of mobility aids (AFMA): Below-knee CASDKAM testing and evaluation program results, J. Rehabil. Res. Dev., vol. 29, no. 4, pp. 78-124, 1992. I?. Kohler, L. Lindh, and P. Netz, “Comparison of CAD/CAM and handmade sockets for PTB prostheses,” Prosthet. Orthot. Znt., vol. 13, no. 1, pp.,, 9-24, 1989. K. E. T. Oberg, “Swedish attempts in using CAD/CAM principles for prosthetics and orthotics,” Clin.-Prosthet. Orthot., vol. 9, pp: 19-23,

1985. K. E. T. Oberg, J. Kofman, A. Karlsson, B. Lindstom, and G. Sigblad, “The CAPOD system-A Scandinavian CAD/CAM system for prosthetic sockets,” J. Prosthet. Orthot., vol. 1, no. 3, pp. 139-148, 1989.

Edward D. Lemaire received the M.Sc, degree in kinanthropology and biomechanics at the University of Ottawa, Ottawa, Canada in 1988 and is currently pursuing the Ph.D degree at the University of Strathclyde’s Bioengineering Unit, Glasgow, Scotland. He is currently a clinical researcher at The Rehabilitation Centre, Ottawa, Canada. His research focus has been on computer applications to prosthetics and orthotics including CAD/CAM, motion analysis, and telemedicine. Mr. Lemaire is a member of the Canadian Societv for Biomechanics and Communications Director for the Canada National Society of the International Society for Prosthetics and Orthotics. He is also the Canadian representative to CAC/ISO/TC1688-WG3 (Prosthetics Testing).

Frank Johnson directs the Medical Engineering Industry Research Affiliation, a joint venture of the University of Ottawa and Carleton University in Canada. The affiliation provides R&D for medical device companies using the expertise of the faculty and research students of the universities.