In current prosthetics practice, prosthetists design sockets and cosmetically shape prostheses on a highly customized basis for every level of amputation. For socket production, a model of the residual limb is made using plaster casting methods; material is added to the model over areas which are pressure intolerant and removed from areas which are loaded for weight-bearing. Both the plaster casting and modification procedures are performed on the basis of skill and judgment and take considerable time.
After the modifications are complete, a socket is fabricated over the mold. To free the socket, the socket is driven off the mold, risking distortion, or the plaster mold is shattered. The latter presents a serious problem. Once the mold is shattered, all shape information is lost. Future fittings of the same amputee usually require the prosthetist to begin the entire process once again. Even if molds are retained, they present a significant storage problem.
Thus, traditional socket manufacturing methods are plagued by inherent difficulties in quantifying and recording the modifica tions used to produce comfortable sockets. Because of this inability to quantify sculpting techniques, expertise is gathered slowly by novice prosthetists and the quality of shape dependent prosthetic components varies from prosthetist to prosthetist. Through the use of computer-aided design and manufacturing (CAD/CAM) technology, many of these problems can be eliminated.
The Computer Aided Socket Design (CASD) program developed by the University of British Columbia Medical Engineering Resource Unit (MERU)1 and Shape Technologies, Inc. has been incorporated into the below-knee module of the CANFIT System, which is an automated shape management system for the prosthetics and orthotics industry.
The system consists of an interactive software package operating on an IBM XT, AT or 386 compatible microcomputer equipped with 640K bytes memory, a hard disk, a mouse, and a high resolution graphics processor and monitor (Figure 1) .
The CANFIT System is designed to be as consistent as possible with current prosthetic practices. The below-knee software module has the capability to develop a "working" socket based on a series of limb measurements. Systematic modifications can be performed on a graphics screen using techniques analogous to rectification of a plaster cast.
Large-scale clinical evaluations of the CANFIT Below-Knee System have been conducted or continue to be underway in five countries: Canada, the Netherlands, Scotland, Sweden, and the United States. During the past three years, over 80 amputees have been successfully fit in 10 centers worldwide. The results of these trials indicate that this automated approach to shape management yields comfortable prostheses and time savings over traditional methods of prosthesis fabrication.
Moreover, recent development work has demonstrated that the CANFIT System has the capability to manage a wide variety of anatomical shapes. The system is currently being expanded to incorporate the design and manufacture of a full range of prosthetics and orthotics, including above-knee sockets, below-knee cosmeses, spinal orthoses, and orthopaedic footwear.
The first step in designing a CANFIT socket is to measure the amputee's residual limb using a tape measure and either standard calipers or the CANFIT Area Tool (Figure 2) . The CANFIT Area Tool is a device used to determine the cross-sectional area of the residuum at various levels. It consists of a rigid front section that approximates the contours of the bony anterior region and a measuring tape that spans the fleshy posterior aspect. Internal functions in the below-knee program convert the area tool measurements to cross-sectional area values.
The CANFIT System is also compatible with more sophisticated devices such as the West Park Research or the Cyberware Below-Knee Laser Scanners. Data from these shape sensing devices can be viewed on the graphics screen and critical measurements can be extracted.
Once the measurements have been entered into the computer, a "working" socket is designed. This is a two-step process in which the program first selects a category of socket shapes from a library of nine reference sockets. These nine sockets, varying in size from small to medium to large, are classified into three groups: tapered, cylindrical, and bulbous, according to the proportion of tissue in the distal portion of the residuum.
The program scales the closest socket shapes in the library both transversely and longitudinally and truncates them to the correct length. To achieve the input measurements, these shapes are then blended by the program to produce a "working" socket shape. By interpolating the "working" shape from a matrix of shapes rather than from a single reference shape, the program ensures that the socket shapes generated by the computer retain critical contours in spite of the sizing process.
One of the difficulties in working with socket reference shapes is the lack of agreement among prosthetists about the contouring of sockets. In dealing with evaluations in North America and Europe, there has been controversy over the shape of the patella tendon, tibial crest, and posterior flare regions. A review of the modifications performed in the various clinical trials indicates that prosthetists with similar backgrounds will consistently try to modify the socket shapes produced by the CANFIT program to be more similar to their conventional hand-made sockets.
To resolve this issue, a method has been established by which the reference shapes can be customized to suit a particular facility. Modified reference shapes have been sent to MERU for measurement on the Shape Copier and subsequent analysis. Various utility programs allow the CANFIT developers to verify the orientation of the redesigned shapes, smooth any digitizing discrepancies, and extract the rear flare from the rest of the shape. These capabilities have allowed customization of the system for clinical centers in Toronto and Holland.
The "working" socket designed by the below-knee program is displayed on a color graphics screen. The display consists of a longitudinal and a transverse cross-sectional view of the socket. The location of each cross-section is selected from an on-screen menu.
Selection of the "Tape Measures" menu item allows for the display of the diameter, circumference, area, and volume of the "working" shape. A hidden-line three dimensional display is also available.
An important feature of the program is its capability to simultaneously display two sockets. This is extremely useful as it allows for comparison of different socket shapes, which are typically "unmodified" and "modified" versions of a single socket.
The CANFIT System enables a prosthetist to modify a working socket to his exact specifications on the basis of previously recorded clinical data, or in an interactive manner according to feedback from the amputee and the expertise of the prosthetist. Three types of modifications are available: patching, size adjustments, and rear flare repositioning. The prosthetist's interactive sculpting tools consist of an on-screen cursor and a mouse which controls the cursor movement.
The patch modification is designed to add or remove relief from certain areas of the socket. This process is analogous to the addition or removal of plaster in strategic areas of a cast. The patch modification is typically used to accommodate bony prominences, generate more load-bearing area, or to increase compression above the femoral condyles. A prosthetist begins by selecting the area to be modified from an on-screen menu. The program then uses the specifications he provides to blend a patch smoothly into the surrounding area.
A prosthetist has control over a wide variety of patching parameters, allowing him to produce virtually any size or shape of patch imaginable. An experienced user will typically fine tune the patching parameters and preview the results of each change until the exact shape of patch desired is achieved. Before any modification is made permanent, it can be previewed on the graphics screen (Figures 3 and Figure 4 ).
For adjustments in size, two modifications are available. The "ply" modification allows the prosthetist to add or remove a specified thickness of material around the entire socket, while the "length" modification permits the length of the socket to be increased or decreased. The rear flare modification treats the entire rear flare area as a separate entity. During the design of the working socket, the program automatically positions the rear flare according to the length of the socket. The rear flare modification allows a prosthetist to override this automatic positioning by enabling him to "pick up" the rear flare and reposition it either longitudinally or in an anterior-posterior direction.
The CANFIT System has been developed to be as user-friendly as possible, allowing those who are unfamiliar with computers to quickly learn to operate the system. In addition, the common problem of searching through cluttered MS-DOS directories has been eliminated by incorporating a built-in "filing" system which stores all client information in an easily accessible form.
Each client file includes clinical informa tion, measurements, and a record of all sockets and modifications performed. Typically, a "working" socket and several subsequent sockets will be stored in each client's file directory. A memo feature allows each socket to be tagged with a short memo or name to enable a prosthetist to easily identify it.
Once a socket shape has been designed, a positive mold of the socket can be produced using a computer numerically-controlled (CNC) milling machine. A manufacturing software program transforms the socket shape into a code which guides a ball nose cutter in the carving of a three-dimensional socket replica from a poly-urethane foam plug (Figure 5) . The result is a very smooth positive mold which does not require hand finishing and may be used to vacuum form or laminate the socket (Figure 6) .
Socket designs may be carved on-site or via modem to a central fabrication facility. There, a socket shape can be carved to specification and returned by courier to the design facility. Within North America, turn-around times of less than 24 hours have been attained.
There are presently four centers in the world which are milling CANFIT socket shapes. Due to the growing number of manufacturing facilities, the CANFIT milling software has been upgraded to accommodate a wider variety of milling machines.
Clinical evaluations of the CANFIT System have been or continue to be carried out in five countries: Canada, the Netherlands, Scotland, Sweden, and the United States. Although numerous centers have been involved, the following have been the most active in conducting trials involving more than 10 subjects:
Het Product Centrum, Tno, Delft, The Netherlands
A fully independent non-profit research organization.
ST. Maartenskliniek, Nijmegen, The Netherlands
An independent orthopaedic hospital.
National Centre for Training in Prosthetics and Orthotics, Strathclyde University, Glasgow, Scotland
An educational and research facility.
Centre for Studies in Aging, Sunny-brook Medical Centre, Toronto, Canada
A research, development, and scientific evaluation center.
While it was considered beneficial to have all the centers perform identical clinical evaluations with identical protocols, this proved impossible to coordinate.
The objectives of the clinical trials varied from center to center. TNO aimed to provide each patient with a socket that was comfortable after one week of use. St. Maartenskliniek provided its patients with permanent CANFIT prostheses. Strathelyde conducted a thorough clinical trial of the CAN-FIT System that included reviewing the shapes generated by the system, critiquing the system for ease of use, and documenting results and observations for feedback to the developers. West Park Research set out to test a scientifically rigorous protocol and estimate the sample size required for a statistically significant study.
In each of the four evaluations, the subjects selected for participation were unilateral amputees with stable residual limbs. All had been using conventionally fabricated limbs for a minimum of three months without problems and were required to walk without walking aids. The subjects included men and women who ranged in age from 15 to 79, and whose surgery had occurred 1 to 43 years prior to the trial. Pathologies associated with the amputations included trauma, vascular deficiency, spinal bifida, and sarcoma.
Prior to beginning the evaluations, each prosthetist participated in an intensive CANFIT training course. The level of conventional prosthetist experience varied between the practitioners as did their familiarity with computers. However, none had significant computer experience.
Fitting procedures also differed. In the case of TNO and St. Maartenskliniek, prosthetists fit sockets consisting of laminated outer shells and soft Pelite liners. At West Park, hardshell sockets were fit with one-ply cotton socks, whereas at Strathclyde, laminated shells with no liners were used.
The results of the four evaluations described above have been summarized in Table 1. Note that a distinction has been made between fittings attempted without liners (for high sensitivity in terms of identifying problems with fit) and those with liners (as commonly used in conventional practice). Separate columns are provided for the average number of fittings required to achieve a satisfactory fit, and the corresponding average number of iterations. This complies with an industry trend in which the first fitting is not included as an iteration. Over 93% of the subjects fit with Pelite lined sockets were fit satisfactorily within 1.9 attempts (0.9 iterations). In the cases which were attempted with hard sockets and no liners, 78% were fit satisfactorily within 2.4 attempts (1.4 iterations).
More than ten other facilities in Europe and North America have been involved in testing the CANFIT System on a limited number of patients from their regular clinical caseloads. One clinic, which has performed independent CANFIT fittings, reported success in six out of six cases, with 1.2 being the average number of fitting attempts. In this case, a single prosthetist trained on the system conducted all of the fittings using laminated sockets with Pelite liners.
The rate of success achieved when fitting CANFIT sockets appears to be governed by several factors. Clearly, in using the CAN-FIT System, sockets with liners are more easily fit than sockets which do not incorporate a liner. This is not surprising as this phenomena is also common in conventional prosthetics technique.
Other observations were as follows:
Amputees with conventional skeletal anatomy in their residual limbs are easier to fit than those with curved tibias or missing bony anatomy.
Proper training is essential for effective use of the system.
Prosthetists obtain successful fittings more quickly once they develop more experience with the CANFIT System. Hence, future users should anticipate a learning curve when they begin working with the system.
The evaluation centers offered three common recommendations to the developers for improving the CANFIT Design software:
The measurement technique should be improved.
More reference shapes should be included in the system.
A capability to customize the reference shapes should be provided.
Each of these recommendations has been addressed in version 3.0 of the CANFIT software, which was released in the summer of 1988. The CANFIT Area Tool has been developed to give more information about the shape and volume of the residuum. A nine reference shape library has been developed, providing for a greater variety of socket shapes and customization of the reference shapes is now available.
Work is underway to expand the capabilities of the CANFIT System to allow for the design and modification of a full range of prosthetic and orthotic shapes. Current development work centers on above-knee sockets, below-knee cosmeses, spinal orthoses, and orthopaedic footwear.
The above-knee program will be, at least outwardly, very similar to the below-knee program. The present prototype software program generates a working socket from a series of measurements and a library of 27 Berkeley-brim (quadrilateral) style reference shapes. Clinical testing of the program has demonstrated the feasibility of the system, while also is targeting areas for improvement. The objective of the below-knee cosmeses program is to generate a stylized shank shape which blends into the existing socket on the basis of several alignment parameters measured with a 3-D digitizer.
The feasibility of developing a CANFIT spinal orthoses program has also been demonstrated. A torso cast was digitized and the resulting measurements were entered into the CANFIT below-knee program. By using the software's patching capabilities, a modified cast suitable for a Milwaukee brace was produced. Plans are currently underway to develop the measurement, design, and modification components for a spinal orthoses software module.
Also under development is an orthopaedic footwear program which will generate a working shoe. The program provides a novel approach to the measurement of feet and the design of shoe lasts.3 A limited number of landmarks on the foot are measured with a 3D digitizer. Based on these measurements, the computer generates a "working" shoe last which can be interactively modified to suit specific requirements (Figure 7) .
The ultimate objective of these development efforts is to incorporate all shape measurement, design, modification, and manufacturing issues that may be encountered in the prosthetics and orthotics industry into a single CANFIT System. This will extend the advantages of CAD/CAM technology-consistency of quality, time savings, increased productivity, and objective record keeping of shapes and shape changes-to all aspects of the prosthetics and orthotics profession.
The generous support of the Workers' Compensation Board of British Columbia and Health and Welfare Canada is gratefully acknowledged.
Carl G. Saunders, M.A.Sc., is the Acting Director of the Medical Engineering Resource Unit at the University of British Columbia and the President of Shape Technologies Inc. Enquiries may be directed to either Shape Technologies Inc., 845 Ruckle Court, North Vancouver, B.C., Canada V7H 2S6, or the Medical Engineering Resource Unit, University Hospital, Shaughnessy Site, 4500 Oak Street, Vancouver, B.C. Canada, V6H 3N I.