109
Technical evaluation of a CAD system for orthopaedic shoe-upper design M Lord, BSc, MPhil, CEng, MBES and J Foulston, BA, MSc, MBES Department of Medical Engineering and Physics, King’s College School of Medicine and Dentistry, London and P J Smith Clarks Shoes Limited, Street, Somerset Computer aided design is now employed routinely in the volume shoe trade. New styles are developed on a three-dimensional image of the last followed by automated pattern generation and engineering. It is suggested that such systems could be useful in the orthopaedic footwear industry although the different requirements for these bespoke products need careful consideration. A clinical trial has been conducted on the Shoemaster (Clarks Shoes) upper design system both to assess its technical capabilities and to consider its role in improvement of service and cosmetic appearance. This particular system works throughout on a three-dimensional representation of the shoe last, which offers particular advantages for integration with shape capture and reproduction. The report concentrates on the technical evaluation to assess (a) its ability to work with unusual last shapes dictated by medical requirements and (b) its potential for integration into a complete computer system for design of both shoe lasts and shoe uppers. The trial indicates that this particular system is promising in both respects. 1 INTRODUCTION
The orthopaedic shoe industry routinely makes bespoke shoes for individuals who are unable to wear normal shoes because of a medical condition. Under the Health Service provision in the United Kingdom, the orthotists or shoe fitters of orthopaedic footwear companies are contracted to visit clinics at hospitals, special schools and residential homes where a consultant doctor can prescribe such footwear for appropriate cases. The footwear is then provided by the orthopaedic company under the supervision of the consultant. An estimated 70000 pairs of bespoke shoes were made in 1990 in England, extrapolated from an informal survey of several major companies. Although detailed costings are not kept by the Department of Health, the annual cost of provision, repair and adaptation of footwear is estimated at 215.6 million in 1988(1). Manufacture at present is largely a craft process in the United Kingdom. For each patient an individual pair of shoe lasts are developed either from a set of measures of the feet or, for more serious deformities, from plaster casts. These shoe lasts are the basic form over which the shoe will be fabricated. Typically the next stage in the process of shoe manufacture requires the development of pattern pieces for the proposed shoe style. A pattern is generated by presenting paper up to the last, and the pattern is then transferred to leather. The leather uppers and liners are cut and ‘closed’ (stitched together) and the entire upper is pulled over the last and attached to the sole unit with appropriate stiffeners and insoles. These practices may vary both in the United Kingdom and abroad. In the United Kingdom the orthotists take the measures and casts, which are then usually relayed back to the central manufacturing facility where last and patterns are generated and the shoes fabricated. In the Netherlands, for example, it is common for the shoemaker to measure up The M S was received on 24 January 1991 and was accepted for publication on 31 July 1991.
H00591 @ IMechE 1991
and make his own last, but to contract the pattern cutting to a specialist. Other models might have the individual shoemaker completing the entire process. In recent years, some dissatisfaction has been expressed with the provision of orthopaedic footwear in the United Kingdom. Surveys of patients revealed a need for improved cosmesis and styling (24). Our own survey of prescribing consultants (5) and a more recent patient questionnaire assessment (6) indicate that the delay in provision, usually taking at least eight weeks from first visit to delivery and often more, is an additional factor. The level of satisfaction varied around the country, with satisfaction with the speed of service ranging from 25 to 67 per cent of responding consultants in the worst and best health service regions. With the craft technology, control of these factors is difficult because of the heavy dependence on skills and manpower in quite small companies. Computer aided technology has been introduced in other manufacturing areas for its advantages in speed and consistency, and graphical systems are now common for aesthetic design. Employment of computer-based technology is then an obvious consideration for solution of some of the perceived needs in orthopaedic footwear provision. Various parts of the manufacturing process are receiving attention as possible contenders for computer aided design (CAD) and manufacture. An overview of the issues presented by introduction of such technology into the field of prosthetics and orthotics is given in Lord and Jones (7). Advanced technology is already widely used in the volume shoe trade (8), and it is an obvious step to investigate the applicability of the available techniques from this sector to the smaller industry of orthopaedic footwear. In the volume trade, CAD has proved itself valuable, particularly in the development of new styles and patterns. The designer draws new styles onto the computer screen. From these images, patterns can be generated automatically for each size in the range and then output directly to a cutter to produce the leather pieces. Many of the systems in use largely work in two dimensions.
0954-4119/91$2.00+ .05 Downloaded from pih.sagepub.com at UQ Library on July 15, 2015
Proc Instn Mech Engrs Vol 205
110
M LORD. J FOULSTON AND P J SMITH
The latest generation of systems provides some threedimensional capability in varying degrees. The comprehensive three-dimensional upper design system evaluated in this report is described more fully below. Although pattern generation is an essential stage also for orthopaedic shoe making, the applicability of computer aided systems developed for the volume trade to bespoke shoe design is not immediately obvious. The primary advantages to the volume trade are speed of development of new style ranges, ability to grade automatically to different widths and lengths, and integration into the production process, for example, in specification of stitching patterns, efficient material layout, planning and costing. These are less relevant, if at all, to one-off bespoke manufacture. However, it is probable that the use of a style library could have applications, allowing well-designed styles to be transferred for the individual case. In isolation, the upper design system might have limited value, but benefits of CAD begin to accrue when more of the production process is linked together. Indeed, a complete system which begins with a digital shape scan of the feet and produces a pair of bespoke lasts with the matching uppers would be most attractive. The consideration of any component part of the system should therefore take into account the potential for integration. An integrated foot measurement, last and pattern design system is probably best implemented on a three-dimensional system. As a first step towards systematic introduction of CAD, this study evaluates an existing three-dimensional CAD shoe-upper system to assess: (a) the ability to work with unusual last shapes dictated by medical requirements and (b) the potential to integrate into a complete computer system for design of both shoe lasts and shoe uppers. 2 DESCRIPTION OF THE CAD DESIGN SYSTEM
A limited clinical trial has been conducted on a threedimensional design system, Shoemaster from Clarks Shoes, to allow both of these technical aspects and service-related factors to be investigated. At the heart of this system is a model of the shoe last. The last shape is initially input into the computer by a method of hand digitization. The last is marked up with a structured grid (Fig. la); one of the lines runs along the featheredge which delimits the upper from the sole. The intersection points of the grid are then fed into the computer (Fig. lb). From this set of surface coordinates, two parametric surface functions are generated for the upper and sole respectively. The limited set of coefficients for this parametric function completely define the surface at all points. Parametric techniques are well documented, for example see Newman and Sproull(9). A t the start of a design session, the model of the last is displayed as a shaded surface on the computer workstation (Fig. 2a) (Unix-based Hewlett Packard). A wireframe representation can also be displayed. A sole unit can be added from a library. To develop the shoe style, style lines are drawn on the three-dimensional image. A wide selection of line styles are available to indicate stitching, cutting and punching. In Fig. 2b the inclusion
(b) Fig. 1 In the last digitization process a shoe last is (a) marked up with a predetermined grid and then (b) digitized manually to a high accuracy
of decorative stitching is shown. When sufficient style lines have been drawn to define the shoe outline and pattern, colour can be specified. The last is then ‘removed’ to give a realistic view of the finished shoe (Fig. 2c). At any point in the design process, the corresponding flattened pattern pieces can be displayed. It is sometimes easier to add certain features onto the twodimensional view, such as circular patterns or straight lines, and then to convert back to three-dimensions. Flattening is the result of a complex algorithm which takes account of the way in which the flat leather pieces will eventually need to stretch around the last during shoe fabrication. At this stage the designer hands over to the pattern engineer, who works with the two-dimensional view to add lasting allowances and optimize pattern layout with the special facilities of the software package, and then the pattern is output to a choice of numerically controlled cutters. In Fig. 3, the pattern pieces have been cut and closed, and the finished shoe is shown fitted to the foot. The system has two important software features in addition to the basic design application, namely grading and transfer. Grading enables the generation of different sizes of last from a single stored last model; in Shoemaster grading is done on the three-dimensional model and each model is flattened to generate the correctly 0 IMechE 1991
Part H ‘ Journal of Engineering in Medicine Downloaded from pih.sagepub.com at UQ Library on July 15, 2015
@ IMechE 1991
N
w r&
. I
111
Proc Instn Mech Engrs Vol 205
TECHNICAL EVALUATION OF A CAD SYSTEM FOR ORTHOPAEDIC SHOE-UPPER DESIGN
Downloaded from pih.sagepub.com at UQ Library on July 15, 2015
M LORD, J FOULSTON AND P J SMITH
112
Fig. 3 The closed upper is shown corresponding to the design in Fig. 1. The completed shoe is fitted to the
Fig. 4 This adult male diabetic patient had undergone forefoot amputation due to vascular insufficiency. His CAD shoes fitted well and minimized his deformity
patient sized pattern pieces. This method is considered to produce a better result than two-dimensional grading of the flattened patterns. The transfer facility enables mapping of an existing style from one last to another of similar size; this is largely automatic and any remaining fine tuning can be done with the interactive graphical tools provided for basic design work. Last shapes can be output to a conventional five-axis CNC mill, a facility that is normally used to generate a range of models of different sizes from a new model. 3 TRIAL PROCEDURE
The clinical trial involved the manufacture and supply of shoes to four adults with diabetic foot problems and six children with severe orthopaedic foot deformities, mostly arising from spina bifida. In the case of two of the adults, the feet were not of particularly abnormal shape but extra clearance was required over the forefoot and extra depth needed to incorporate special inserts. Of the remaining adults, one had a forefoot amputation (Fig. 4) and another a large dorsal tuberosity which needed accommodation. The children’s feet were generally short, bulky and outside a normal range of foot shapes: two had a pair of feet of markedly different sizes and two required the use of external orthoses. For example, the patient in Fig. 3 is wearing moulded anklefoot orthoses under her socks. Full details of the clinical aspects of the trial are published elsewhere (10).
The established orthopaedic lasts for these patients were digitized with any cradle or special insert attached. To aid digitization, a thin drape was pulled over the rough last surface to present a uniform hard surface. A design session was arranged at the location of each orthopaedic company to which patients, orthotists, lastmakers, two experienced shoe designers and the research coordinators convened. The designers worked with the patient and relatives at the CAD station to achieve a style acceptable to the patient. Styles were generated either as complete originals (seven pairs) or by transfer of current Clarks’ styles which had previously been identified as suitable (three pairs). In each case, the design was completed over one last and then transferred to the other last. In cases where the feet were of substantially different size, the design was first graded in three-dimensions to an appropriate size and then transferred. Subsequently the shoe uppers were cut and closed at Clarks Shoe factory, and lasted and fitted by one of the two orthopaedic companies involved in the trial. A repeat order requested by a clinician for one patient was outside the protocol of this trial. However, a paper pattern was supplied from the CAD system and used in the conventional way at the orthopaedic company to manufacture a new pair of shoes. Both leather and lining patterns were supplied. All procedures were closely observed by the company orthotist, the last-maker, the design engineer and the research team. Technical aspects of the procedure,
Part H : Journal of Engineering in Medicine
@ IMechE 1991 Downloaded from pih.sagepub.com at UQ Library on July 15, 2015
TECHNICAL EVALUATION OF A CAD SYSTEM FOR ORTHOPAEDIC SHOE-UPPER DESIGN
113
including any difficulties encountered, were noted alongside the clinical aspects. 4 RESULTS
Overall the CAD system proved versatile and able to meet the orthopaedic requirements. The shoes ranged from lace-up Derby styles through to trainer shoes and T-bars for the children. No insurmountable problems were encountered with the lasting and fitting of these shoes. Only one pair of shoes failed to fit due to inadequate depth to accommodate the cradle in the heel area; this was essentially a problem introduced a t the time of lasting. Although it is not of direct relevance to this technical report, it can be noted that the trial shoes were well received in terms of their improved cosmesis and high-quality leathers. 4.1 Balancing of design When the designs at first produced for one shoe were mapped onto the other last of the pair, the designers spent some time to achieve the correct cosmetic ‘balancing’. This process is not needed with the normal symmetrical pair, and the CAD package does not have the facility to display the right and left last at the same time during design for comparison. This made the process longer than would otherwise be necessary. By maintaining each shoe roughly in proportion, a very good result was achieved. The transfer of patterns did not result in any distortions of, for example, decorative stitching and their placement, and the proportions of the various parts of the shoes were maintained in balance.
Fig. 5
A discrepancy can be noted between the closing of the lace panel on the three-dimensional image and the fitted shoe for the bulkier right foot, although this did not prevent a satisfactory fit
4.2 Symmetry of design
With original designs, the normal practice of reflecting the design about a straight line down the middle of the last could not always be practised. This is due to the unusual shapes of the lasts. Nevertheless, the CAD system allows the option to develop each side separately, and this was therefore not a problem. 4.3 Design problems from high mid-foot
noted in passing that the sole unit fits exactly around the base of the custom last, rather than appearing as a standard size and width, so that the individual last shapes presented no problems. The shoes were eventually lasted and the soles provided at the orthopaedic company, where special features such as heel flare were occasionally required. Printouts of the screen design were not made available at the time of lasting, and the difference between the sole unit on the screen printout and that supplied in practice could influence the cosmetic appearance considerably.
Two of the children’s shoes with a T-bar tended to gape on the upright of the ‘T’,which can be seen by close inspection of the right shoe in Fig. 3. Problems were also experienced in this area with lace-up shoes, where the panel was held too wide open (e.g. Fig. 5). In all cases, this occurred where there was considerable foot deformity in the mid-foot with increased girth and height. It was noted from screen copies that the problem did not appear on the three-dimensional image, indicating that the problem occurred in the flattening stage. The flattening algorithm is sensitive in this area, and future software development is indicated, although this was not a severe problem. Indeed, the research team noted the gaping, but no patient commented on this.
The designer draws onto a representation of the last, which in the orthopaedic case included any special inserts and allowances. This leads to confusion as to what is space for the foot and what is space for the insert. If the designer draws a top line in what appears to be the normal position, then this may result in a shallow shoe around the heel and under the ankles when the insert is in place. In practice the seam allowances were sufficient to allow for correction at the time of lasting, but resulted in tight working.
4.4 Sole units During design, an arbitrary sole unit was selected from a library of such units available at Clarks Shoes. It was
4.6 Paper patterns Paper patterns were found to be useful for the production of a second pair of shoes for an adult patient.
4.5 Back seam and under-ankle heights
Proc Instn Mech Engrs Vol 205
@ IMechE 1991 Downloaded from pih.sagepub.com at UQ Library on July 15, 2015
M LORD, J FOULSTON AND P J SMITH
I14
The computer system generated both upper and liner pieces, with the liners being slightly smaller. It was not the normal practice in the orthopaedic company to use different patterns due to time involved in their production, but the orthotist considers that this gives a superior result and should be used if the CAD system can automatically generate liner patterns with no time overhead.
paedic footwear, and to communication of this information to the place of lasting and sole attachment. There is obvious need for improvement of the flattening algorithm in the mid-foot region where this is substantially different from normal. More experimentation is required to establish the particular conditions that gave rise to inaccuracies over the instep. The ability of the system to produce low boots is under consideration.
5 DISCUSSION
5.1 Overall performance
5.3 Integration
The results of this evaluation are highly encouraging. Initial doubts as to the capabilities of a CAD system to operate away from the symmetry of the standard application, or to cope with shapes substantially different from those of normal lasts, have been dispelled. Obviously this statement must be qualified as to the limited range of last shapes that have been subject to trial, comprising those that can normally be dealt with by adaptation of a standard last. Further tests are indicated to confirm the capabilities for gross deformity where the last-maker would normally work from a plaster-cast of the foot.
Although for the purpose of this trial the starting point was a hand-crafted last, this would not be a viable production route. The vital stages in a fully integrated CAD system are foot shape capture and CAD last design, followed by CAD upper design and CNC production of lasts. It is appropriate at this stage to review the potential of Shoemaster within this integration. Optical methods of foot shape capture are under development, for example in the system developed at LIC Orthopaedic in Sweden which can capture the x y z coordinates of tens of thousands of points from the surface of a live foot in a matter of seconds. Conversion of foot shape data into an appropriate last shape can be approached in two ways, either by transformation of the foot shape into last shape (by some algorithm as yet unknown) or by matching the foot to the best available last from a predigitized library of lasts. In the latter method, which must be the method of choice at present, a library of shoe lasts needs to be developed. This in itself requires digitization from models. The hand-held stylus digitization method used for this trial is time consuming, but has the advantage of producing a grid of structured data to which the surface model can be fitted. Optical scanning is more rapid, but produces unstructured surface coordinates. This unstructured data must be reduced into a structured form for conversion into the parametric representation central to the Shoemaster system. In this task, it is necessary to identify very accurately the feather edge, which is a sharp transition between the upper and sole in the fore-foot region, and to parameterize the two surfaces correctly. The development of such algorithms should be a priority for software engineers wishing to capture last shapes from an optical scan for input into this type of patterngenerating software. Work in this area is already reported (11). With either library lasts or one-off lasts, a threedimensional on-screen sculpting facility is probably required for fine tuning or adjustment of the last. One might even envisage the development of new last shapes from scratch via computer graphics. The parametric representation in the package under evaluation is ideal for this use. Such parametric representations were originally developed for the design of car bodies where modification of a doubly curved surface is also required. Development of a facility to design and modify last shapes on-screen is already projected at Shoemaster. At the end of the process, the CNC machine currently interfaced with the CAD system is a standard industrial mill which takes several hours to produce a model last. Such lasts are of high accuracy and good finish. The model last is then transferred to a traditional last-
5.2 The design system Both the transfer and the grading facilities were essential parts of the operation, although used in a different capacity from their normal function. In volume design, these facilities allow ranges of sizes to be developed over slightly different fashion shapes. In this orthopaedic trial a combination of grading and transfer was used to map existing styles onto the custom last and to map original designs for odd-sized feet. The success of this operation is attributable to the ability to grade in three dimensions. The operation could not be achieved in this way on a two-dimensional system, where grading is done on the flat pattern pieces. The design process is inextricably linked to the availability of certain leathers, stitching techniques, accessories and punching which can be specified on the workstation. In this trial both the upper design and the cutting and closing of the uppers were done by the same team, that is at Shoemaster. Development of extra facilities for orthopaedic usage are indicated in : Balancing. A facility to display both shoes on-screen during styling is needed to achieve the best balance in the design where the feet are of markedly different dimensions. Line of insert. The designer needs an indication on the three-dimensional image of the location of any insert to prevent a faulty perception of where the foot is in relation to the style lines being drawn. Indication of heel backseam and under-ankle height. These heights should be specified from patient measures, and a facility added to ensure that the top line passes through the appropriate points. Perhaps the line of the insert or insert allowance could be indicated on the last model. Sole units. Consideration should be given to addition to the library of typical soles provided by the ortho-
@ IMechE 1991
Part H : Journal of Engineering in Medicine Downloaded from pih.sagepub.com at UQ Library on July 15, 2015
TECHNICAL EVALUATION O F A CAD SYSTEM FOR ORTHOPAEDIC SHOE-UPPER DESIGN
copying machine which can turn out a production copy in under two minutes. An alternative lower-cost method referred to by Duncan and Mair (12) as polyhedral machining is employed elsewhere (13); this method, involving the slow rotation of a blank in a lathe while the milling tool moves radially and slowly traverses in a spiral path along the blank, has been adopted commercially for the generation of moulds for prosthetic sockets in a similar application. However, the process is still fairly slow, over 20 minutes per model, and the surface finish and accuracy is less than that demanded in commercial shoe work. Fortunately, more than one commercial company has already demonstrated it is possible to retro-fit the traditional last-copier with a CNC input stage and to produce custom lasts to high commercial standards in a reasonable time of under ten minutes. Preliminary discussions have confirmed the compatibility between the output of the CAD system and such a CNC last-making machine. Commercial availability should ensue.
5.4 Service implications In service, it would not be feasible to consider making original designs for individual patients. This system has shown that styles held in a style library can be mapped over the individual three-dimensional last as required and in a reasonable time. Further trials are indicated to assess realistic timings for this process. A number of the current high-street styles were found to be suitable in the children’s ranges and the exciting prospect of providing fashionable shoes is a real option.
6 CONCLUSION
The results of this trial of a commercial computer aided design system are encouraging for its potential use with orthopaedic shoes. Special features of the system can be adapted to meet the one-off nature of this application. An evaluation of the system shows its potential for integration into a fully automated system for design of both shoe last and shoe uppers, with projected extensions for the volume trade which would also be of benefit for orthopaedic use.
1 I5
ACKNOWLEDGEMENTS
This project was partly supported by a development grant from the British Diabetic Association with medical collaboration from Dr M. E. Edmonds of King’s College Hospital, London. The authors acknowledge the considerable contributions of design engineers H. Bishop, E. Duffey and B. Holley of Clarks Shoes Limited, and contributions of B. Holback and S. Lloyd of LSB Orthopaedics Limited and C. Peacock and P. Charlton of J. C. Peacock and Son Limited (director and senior orthotist of each company respectively). The authors particularly thank the Directors of Clarks Shoes for permission to use the Clarks styles. REFERENCES 1 Study ofthe Orthotic Service. NHS Management Consultancy Services, 1988. 2 Office of Population Censuses and Surveys. National Health Service surgical footwear. A study of patient sutisfacrion, 1979 (HMSO, London). 3 Guthrie, D. A future for the orthotic service: a report by the working party on the availability and supply of orthotic appliances, 1983, The Royal Association for Disability and Rehabilitation, London. 4 Costigan, P. S., Miller, G., Elliot, C. and Wallace, W. A. Are surgical shoes providing value for money? Br. M e d . J . , 1989, 299, 950. 5 Lord, M. and Foulston, J. Surgical footwear: a survey of prescribing consultants. Br. Med. J . , 1989,299, 657. 6 Fisher, L. R. and McLellan, D. L. Questionnaire assessment of patient satisfaction with lower limb orthoses from a district hospital. Prosthetics and Orthotics Int., 1989, 13, 29-35. 7 Lord, M. and Jones, D. Issues and themes in computer aided design for external prosthetics and orthotics. J . Biomed. Enyng, 1988,10,491498. 8 Smith, P. J. Shoe design and manufacture by computer. Con,ference on Design engineering, Birmingham, 1984 (Institution of Mechanical Engineers). 9 Newman, W. M. and Spronll, R. F. Principles of inteructiw computer graphics, 1981 (McGraw-Hill). 10 Lord, M. and Foulston, J. Clinical trial of a computer-aided design system for oathopaedic shoe upper design. Prosthetics and Orthotics Int., 1991, 15, 11-17. 11 Lord, M. and Travis, R. P. Surface modelling the foot from OSIRIS scans. In Surface topography and body deformity (Eds H. Neugebauer and G. Windischbauer), 1990 (Gustav Fischer Verlag). 12 Duncan, J. P. and Mair, S. G. Sculptured surfaces in engineering and medicine, 1983 (Cambridge University Press). 13 Saunders, C. G. and Vickers, G. W. A generalised approach to the replication of cylindrical bodies with compound curvature. Trans. A S M E , 1984,106,7&76.
Proc Instn Mech Engrs Vol 205
@ IMechE 1991 Downloaded from pih.sagepub.com at UQ Library on July 15, 2015
Technical evaluation of a CAD system for orthopaedic shoe-upper design M Lord, BSc, MPhil, CEng, MBES and J Foulston, BA, MSc, MBES Department of Medical Engineering and Physics, King’s College School of Medicine and Dentistry, London and P J Smith Clarks Shoes Limited, Street, Somerset Computer aided design is now employed routinely in the volume shoe trade. New styles are developed on a three-dimensional image of the last followed by automated pattern generation and engineering. It is suggested that such systems could be useful in the orthopaedic footwear industry although the different requirements for these bespoke products need careful consideration. A clinical trial has been conducted on the Shoemaster (Clarks Shoes) upper design system both to assess its technical capabilities and to consider its role in improvement of service and cosmetic appearance. This particular system works throughout on a three-dimensional representation of the shoe last, which offers particular advantages for integration with shape capture and reproduction. The report concentrates on the technical evaluation to assess (a) its ability to work with unusual last shapes dictated by medical requirements and (b) its potential for integration into a complete computer system for design of both shoe lasts and shoe uppers. The trial indicates that this particular system is promising in both respects. 1 INTRODUCTION
The orthopaedic shoe industry routinely makes bespoke shoes for individuals who are unable to wear normal shoes because of a medical condition. Under the Health Service provision in the United Kingdom, the orthotists or shoe fitters of orthopaedic footwear companies are contracted to visit clinics at hospitals, special schools and residential homes where a consultant doctor can prescribe such footwear for appropriate cases. The footwear is then provided by the orthopaedic company under the supervision of the consultant. An estimated 70000 pairs of bespoke shoes were made in 1990 in England, extrapolated from an informal survey of several major companies. Although detailed costings are not kept by the Department of Health, the annual cost of provision, repair and adaptation of footwear is estimated at 215.6 million in 1988(1). Manufacture at present is largely a craft process in the United Kingdom. For each patient an individual pair of shoe lasts are developed either from a set of measures of the feet or, for more serious deformities, from plaster casts. These shoe lasts are the basic form over which the shoe will be fabricated. Typically the next stage in the process of shoe manufacture requires the development of pattern pieces for the proposed shoe style. A pattern is generated by presenting paper up to the last, and the pattern is then transferred to leather. The leather uppers and liners are cut and ‘closed’ (stitched together) and the entire upper is pulled over the last and attached to the sole unit with appropriate stiffeners and insoles. These practices may vary both in the United Kingdom and abroad. In the United Kingdom the orthotists take the measures and casts, which are then usually relayed back to the central manufacturing facility where last and patterns are generated and the shoes fabricated. In the Netherlands, for example, it is common for the shoemaker to measure up The M S was received on 24 January 1991 and was accepted for publication on 31 July 1991.
H00591 @ IMechE 1991
and make his own last, but to contract the pattern cutting to a specialist. Other models might have the individual shoemaker completing the entire process. In recent years, some dissatisfaction has been expressed with the provision of orthopaedic footwear in the United Kingdom. Surveys of patients revealed a need for improved cosmesis and styling (24). Our own survey of prescribing consultants (5) and a more recent patient questionnaire assessment (6) indicate that the delay in provision, usually taking at least eight weeks from first visit to delivery and often more, is an additional factor. The level of satisfaction varied around the country, with satisfaction with the speed of service ranging from 25 to 67 per cent of responding consultants in the worst and best health service regions. With the craft technology, control of these factors is difficult because of the heavy dependence on skills and manpower in quite small companies. Computer aided technology has been introduced in other manufacturing areas for its advantages in speed and consistency, and graphical systems are now common for aesthetic design. Employment of computer-based technology is then an obvious consideration for solution of some of the perceived needs in orthopaedic footwear provision. Various parts of the manufacturing process are receiving attention as possible contenders for computer aided design (CAD) and manufacture. An overview of the issues presented by introduction of such technology into the field of prosthetics and orthotics is given in Lord and Jones (7). Advanced technology is already widely used in the volume shoe trade (8), and it is an obvious step to investigate the applicability of the available techniques from this sector to the smaller industry of orthopaedic footwear. In the volume trade, CAD has proved itself valuable, particularly in the development of new styles and patterns. The designer draws new styles onto the computer screen. From these images, patterns can be generated automatically for each size in the range and then output directly to a cutter to produce the leather pieces. Many of the systems in use largely work in two dimensions.
0954-4119/91$2.00+ .05 Downloaded from pih.sagepub.com at UQ Library on July 15, 2015
Proc Instn Mech Engrs Vol 205
110
M LORD. J FOULSTON AND P J SMITH
The latest generation of systems provides some threedimensional capability in varying degrees. The comprehensive three-dimensional upper design system evaluated in this report is described more fully below. Although pattern generation is an essential stage also for orthopaedic shoe making, the applicability of computer aided systems developed for the volume trade to bespoke shoe design is not immediately obvious. The primary advantages to the volume trade are speed of development of new style ranges, ability to grade automatically to different widths and lengths, and integration into the production process, for example, in specification of stitching patterns, efficient material layout, planning and costing. These are less relevant, if at all, to one-off bespoke manufacture. However, it is probable that the use of a style library could have applications, allowing well-designed styles to be transferred for the individual case. In isolation, the upper design system might have limited value, but benefits of CAD begin to accrue when more of the production process is linked together. Indeed, a complete system which begins with a digital shape scan of the feet and produces a pair of bespoke lasts with the matching uppers would be most attractive. The consideration of any component part of the system should therefore take into account the potential for integration. An integrated foot measurement, last and pattern design system is probably best implemented on a three-dimensional system. As a first step towards systematic introduction of CAD, this study evaluates an existing three-dimensional CAD shoe-upper system to assess: (a) the ability to work with unusual last shapes dictated by medical requirements and (b) the potential to integrate into a complete computer system for design of both shoe lasts and shoe uppers. 2 DESCRIPTION OF THE CAD DESIGN SYSTEM
A limited clinical trial has been conducted on a threedimensional design system, Shoemaster from Clarks Shoes, to allow both of these technical aspects and service-related factors to be investigated. At the heart of this system is a model of the shoe last. The last shape is initially input into the computer by a method of hand digitization. The last is marked up with a structured grid (Fig. la); one of the lines runs along the featheredge which delimits the upper from the sole. The intersection points of the grid are then fed into the computer (Fig. lb). From this set of surface coordinates, two parametric surface functions are generated for the upper and sole respectively. The limited set of coefficients for this parametric function completely define the surface at all points. Parametric techniques are well documented, for example see Newman and Sproull(9). A t the start of a design session, the model of the last is displayed as a shaded surface on the computer workstation (Fig. 2a) (Unix-based Hewlett Packard). A wireframe representation can also be displayed. A sole unit can be added from a library. To develop the shoe style, style lines are drawn on the three-dimensional image. A wide selection of line styles are available to indicate stitching, cutting and punching. In Fig. 2b the inclusion
(b) Fig. 1 In the last digitization process a shoe last is (a) marked up with a predetermined grid and then (b) digitized manually to a high accuracy
of decorative stitching is shown. When sufficient style lines have been drawn to define the shoe outline and pattern, colour can be specified. The last is then ‘removed’ to give a realistic view of the finished shoe (Fig. 2c). At any point in the design process, the corresponding flattened pattern pieces can be displayed. It is sometimes easier to add certain features onto the twodimensional view, such as circular patterns or straight lines, and then to convert back to three-dimensions. Flattening is the result of a complex algorithm which takes account of the way in which the flat leather pieces will eventually need to stretch around the last during shoe fabrication. At this stage the designer hands over to the pattern engineer, who works with the two-dimensional view to add lasting allowances and optimize pattern layout with the special facilities of the software package, and then the pattern is output to a choice of numerically controlled cutters. In Fig. 3, the pattern pieces have been cut and closed, and the finished shoe is shown fitted to the foot. The system has two important software features in addition to the basic design application, namely grading and transfer. Grading enables the generation of different sizes of last from a single stored last model; in Shoemaster grading is done on the three-dimensional model and each model is flattened to generate the correctly 0 IMechE 1991
Part H ‘ Journal of Engineering in Medicine Downloaded from pih.sagepub.com at UQ Library on July 15, 2015
@ IMechE 1991
N
w r&
. I
111
Proc Instn Mech Engrs Vol 205
TECHNICAL EVALUATION OF A CAD SYSTEM FOR ORTHOPAEDIC SHOE-UPPER DESIGN
Downloaded from pih.sagepub.com at UQ Library on July 15, 2015
M LORD, J FOULSTON AND P J SMITH
112
Fig. 3 The closed upper is shown corresponding to the design in Fig. 1. The completed shoe is fitted to the
Fig. 4 This adult male diabetic patient had undergone forefoot amputation due to vascular insufficiency. His CAD shoes fitted well and minimized his deformity
patient sized pattern pieces. This method is considered to produce a better result than two-dimensional grading of the flattened patterns. The transfer facility enables mapping of an existing style from one last to another of similar size; this is largely automatic and any remaining fine tuning can be done with the interactive graphical tools provided for basic design work. Last shapes can be output to a conventional five-axis CNC mill, a facility that is normally used to generate a range of models of different sizes from a new model. 3 TRIAL PROCEDURE
The clinical trial involved the manufacture and supply of shoes to four adults with diabetic foot problems and six children with severe orthopaedic foot deformities, mostly arising from spina bifida. In the case of two of the adults, the feet were not of particularly abnormal shape but extra clearance was required over the forefoot and extra depth needed to incorporate special inserts. Of the remaining adults, one had a forefoot amputation (Fig. 4) and another a large dorsal tuberosity which needed accommodation. The children’s feet were generally short, bulky and outside a normal range of foot shapes: two had a pair of feet of markedly different sizes and two required the use of external orthoses. For example, the patient in Fig. 3 is wearing moulded anklefoot orthoses under her socks. Full details of the clinical aspects of the trial are published elsewhere (10).
The established orthopaedic lasts for these patients were digitized with any cradle or special insert attached. To aid digitization, a thin drape was pulled over the rough last surface to present a uniform hard surface. A design session was arranged at the location of each orthopaedic company to which patients, orthotists, lastmakers, two experienced shoe designers and the research coordinators convened. The designers worked with the patient and relatives at the CAD station to achieve a style acceptable to the patient. Styles were generated either as complete originals (seven pairs) or by transfer of current Clarks’ styles which had previously been identified as suitable (three pairs). In each case, the design was completed over one last and then transferred to the other last. In cases where the feet were of substantially different size, the design was first graded in three-dimensions to an appropriate size and then transferred. Subsequently the shoe uppers were cut and closed at Clarks Shoe factory, and lasted and fitted by one of the two orthopaedic companies involved in the trial. A repeat order requested by a clinician for one patient was outside the protocol of this trial. However, a paper pattern was supplied from the CAD system and used in the conventional way at the orthopaedic company to manufacture a new pair of shoes. Both leather and lining patterns were supplied. All procedures were closely observed by the company orthotist, the last-maker, the design engineer and the research team. Technical aspects of the procedure,
Part H : Journal of Engineering in Medicine
@ IMechE 1991 Downloaded from pih.sagepub.com at UQ Library on July 15, 2015
TECHNICAL EVALUATION OF A CAD SYSTEM FOR ORTHOPAEDIC SHOE-UPPER DESIGN
113
including any difficulties encountered, were noted alongside the clinical aspects. 4 RESULTS
Overall the CAD system proved versatile and able to meet the orthopaedic requirements. The shoes ranged from lace-up Derby styles through to trainer shoes and T-bars for the children. No insurmountable problems were encountered with the lasting and fitting of these shoes. Only one pair of shoes failed to fit due to inadequate depth to accommodate the cradle in the heel area; this was essentially a problem introduced a t the time of lasting. Although it is not of direct relevance to this technical report, it can be noted that the trial shoes were well received in terms of their improved cosmesis and high-quality leathers. 4.1 Balancing of design When the designs at first produced for one shoe were mapped onto the other last of the pair, the designers spent some time to achieve the correct cosmetic ‘balancing’. This process is not needed with the normal symmetrical pair, and the CAD package does not have the facility to display the right and left last at the same time during design for comparison. This made the process longer than would otherwise be necessary. By maintaining each shoe roughly in proportion, a very good result was achieved. The transfer of patterns did not result in any distortions of, for example, decorative stitching and their placement, and the proportions of the various parts of the shoes were maintained in balance.
Fig. 5
A discrepancy can be noted between the closing of the lace panel on the three-dimensional image and the fitted shoe for the bulkier right foot, although this did not prevent a satisfactory fit
4.2 Symmetry of design
With original designs, the normal practice of reflecting the design about a straight line down the middle of the last could not always be practised. This is due to the unusual shapes of the lasts. Nevertheless, the CAD system allows the option to develop each side separately, and this was therefore not a problem. 4.3 Design problems from high mid-foot
noted in passing that the sole unit fits exactly around the base of the custom last, rather than appearing as a standard size and width, so that the individual last shapes presented no problems. The shoes were eventually lasted and the soles provided at the orthopaedic company, where special features such as heel flare were occasionally required. Printouts of the screen design were not made available at the time of lasting, and the difference between the sole unit on the screen printout and that supplied in practice could influence the cosmetic appearance considerably.
Two of the children’s shoes with a T-bar tended to gape on the upright of the ‘T’,which can be seen by close inspection of the right shoe in Fig. 3. Problems were also experienced in this area with lace-up shoes, where the panel was held too wide open (e.g. Fig. 5). In all cases, this occurred where there was considerable foot deformity in the mid-foot with increased girth and height. It was noted from screen copies that the problem did not appear on the three-dimensional image, indicating that the problem occurred in the flattening stage. The flattening algorithm is sensitive in this area, and future software development is indicated, although this was not a severe problem. Indeed, the research team noted the gaping, but no patient commented on this.
The designer draws onto a representation of the last, which in the orthopaedic case included any special inserts and allowances. This leads to confusion as to what is space for the foot and what is space for the insert. If the designer draws a top line in what appears to be the normal position, then this may result in a shallow shoe around the heel and under the ankles when the insert is in place. In practice the seam allowances were sufficient to allow for correction at the time of lasting, but resulted in tight working.
4.4 Sole units During design, an arbitrary sole unit was selected from a library of such units available at Clarks Shoes. It was
4.6 Paper patterns Paper patterns were found to be useful for the production of a second pair of shoes for an adult patient.
4.5 Back seam and under-ankle heights
Proc Instn Mech Engrs Vol 205
@ IMechE 1991 Downloaded from pih.sagepub.com at UQ Library on July 15, 2015
M LORD, J FOULSTON AND P J SMITH
I14
The computer system generated both upper and liner pieces, with the liners being slightly smaller. It was not the normal practice in the orthopaedic company to use different patterns due to time involved in their production, but the orthotist considers that this gives a superior result and should be used if the CAD system can automatically generate liner patterns with no time overhead.
paedic footwear, and to communication of this information to the place of lasting and sole attachment. There is obvious need for improvement of the flattening algorithm in the mid-foot region where this is substantially different from normal. More experimentation is required to establish the particular conditions that gave rise to inaccuracies over the instep. The ability of the system to produce low boots is under consideration.
5 DISCUSSION
5.1 Overall performance
5.3 Integration
The results of this evaluation are highly encouraging. Initial doubts as to the capabilities of a CAD system to operate away from the symmetry of the standard application, or to cope with shapes substantially different from those of normal lasts, have been dispelled. Obviously this statement must be qualified as to the limited range of last shapes that have been subject to trial, comprising those that can normally be dealt with by adaptation of a standard last. Further tests are indicated to confirm the capabilities for gross deformity where the last-maker would normally work from a plaster-cast of the foot.
Although for the purpose of this trial the starting point was a hand-crafted last, this would not be a viable production route. The vital stages in a fully integrated CAD system are foot shape capture and CAD last design, followed by CAD upper design and CNC production of lasts. It is appropriate at this stage to review the potential of Shoemaster within this integration. Optical methods of foot shape capture are under development, for example in the system developed at LIC Orthopaedic in Sweden which can capture the x y z coordinates of tens of thousands of points from the surface of a live foot in a matter of seconds. Conversion of foot shape data into an appropriate last shape can be approached in two ways, either by transformation of the foot shape into last shape (by some algorithm as yet unknown) or by matching the foot to the best available last from a predigitized library of lasts. In the latter method, which must be the method of choice at present, a library of shoe lasts needs to be developed. This in itself requires digitization from models. The hand-held stylus digitization method used for this trial is time consuming, but has the advantage of producing a grid of structured data to which the surface model can be fitted. Optical scanning is more rapid, but produces unstructured surface coordinates. This unstructured data must be reduced into a structured form for conversion into the parametric representation central to the Shoemaster system. In this task, it is necessary to identify very accurately the feather edge, which is a sharp transition between the upper and sole in the fore-foot region, and to parameterize the two surfaces correctly. The development of such algorithms should be a priority for software engineers wishing to capture last shapes from an optical scan for input into this type of patterngenerating software. Work in this area is already reported (11). With either library lasts or one-off lasts, a threedimensional on-screen sculpting facility is probably required for fine tuning or adjustment of the last. One might even envisage the development of new last shapes from scratch via computer graphics. The parametric representation in the package under evaluation is ideal for this use. Such parametric representations were originally developed for the design of car bodies where modification of a doubly curved surface is also required. Development of a facility to design and modify last shapes on-screen is already projected at Shoemaster. At the end of the process, the CNC machine currently interfaced with the CAD system is a standard industrial mill which takes several hours to produce a model last. Such lasts are of high accuracy and good finish. The model last is then transferred to a traditional last-
5.2 The design system Both the transfer and the grading facilities were essential parts of the operation, although used in a different capacity from their normal function. In volume design, these facilities allow ranges of sizes to be developed over slightly different fashion shapes. In this orthopaedic trial a combination of grading and transfer was used to map existing styles onto the custom last and to map original designs for odd-sized feet. The success of this operation is attributable to the ability to grade in three dimensions. The operation could not be achieved in this way on a two-dimensional system, where grading is done on the flat pattern pieces. The design process is inextricably linked to the availability of certain leathers, stitching techniques, accessories and punching which can be specified on the workstation. In this trial both the upper design and the cutting and closing of the uppers were done by the same team, that is at Shoemaster. Development of extra facilities for orthopaedic usage are indicated in : Balancing. A facility to display both shoes on-screen during styling is needed to achieve the best balance in the design where the feet are of markedly different dimensions. Line of insert. The designer needs an indication on the three-dimensional image of the location of any insert to prevent a faulty perception of where the foot is in relation to the style lines being drawn. Indication of heel backseam and under-ankle height. These heights should be specified from patient measures, and a facility added to ensure that the top line passes through the appropriate points. Perhaps the line of the insert or insert allowance could be indicated on the last model. Sole units. Consideration should be given to addition to the library of typical soles provided by the ortho-
@ IMechE 1991
Part H : Journal of Engineering in Medicine Downloaded from pih.sagepub.com at UQ Library on July 15, 2015
TECHNICAL EVALUATION O F A CAD SYSTEM FOR ORTHOPAEDIC SHOE-UPPER DESIGN
copying machine which can turn out a production copy in under two minutes. An alternative lower-cost method referred to by Duncan and Mair (12) as polyhedral machining is employed elsewhere (13); this method, involving the slow rotation of a blank in a lathe while the milling tool moves radially and slowly traverses in a spiral path along the blank, has been adopted commercially for the generation of moulds for prosthetic sockets in a similar application. However, the process is still fairly slow, over 20 minutes per model, and the surface finish and accuracy is less than that demanded in commercial shoe work. Fortunately, more than one commercial company has already demonstrated it is possible to retro-fit the traditional last-copier with a CNC input stage and to produce custom lasts to high commercial standards in a reasonable time of under ten minutes. Preliminary discussions have confirmed the compatibility between the output of the CAD system and such a CNC last-making machine. Commercial availability should ensue.
5.4 Service implications In service, it would not be feasible to consider making original designs for individual patients. This system has shown that styles held in a style library can be mapped over the individual three-dimensional last as required and in a reasonable time. Further trials are indicated to assess realistic timings for this process. A number of the current high-street styles were found to be suitable in the children’s ranges and the exciting prospect of providing fashionable shoes is a real option.
6 CONCLUSION
The results of this trial of a commercial computer aided design system are encouraging for its potential use with orthopaedic shoes. Special features of the system can be adapted to meet the one-off nature of this application. An evaluation of the system shows its potential for integration into a fully automated system for design of both shoe last and shoe uppers, with projected extensions for the volume trade which would also be of benefit for orthopaedic use.
1 I5
ACKNOWLEDGEMENTS
This project was partly supported by a development grant from the British Diabetic Association with medical collaboration from Dr M. E. Edmonds of King’s College Hospital, London. The authors acknowledge the considerable contributions of design engineers H. Bishop, E. Duffey and B. Holley of Clarks Shoes Limited, and contributions of B. Holback and S. Lloyd of LSB Orthopaedics Limited and C. Peacock and P. Charlton of J. C. Peacock and Son Limited (director and senior orthotist of each company respectively). The authors particularly thank the Directors of Clarks Shoes for permission to use the Clarks styles. REFERENCES 1 Study ofthe Orthotic Service. NHS Management Consultancy Services, 1988. 2 Office of Population Censuses and Surveys. National Health Service surgical footwear. A study of patient sutisfacrion, 1979 (HMSO, London). 3 Guthrie, D. A future for the orthotic service: a report by the working party on the availability and supply of orthotic appliances, 1983, The Royal Association for Disability and Rehabilitation, London. 4 Costigan, P. S., Miller, G., Elliot, C. and Wallace, W. A. Are surgical shoes providing value for money? Br. M e d . J . , 1989, 299, 950. 5 Lord, M. and Foulston, J. Surgical footwear: a survey of prescribing consultants. Br. Med. J . , 1989,299, 657. 6 Fisher, L. R. and McLellan, D. L. Questionnaire assessment of patient satisfaction with lower limb orthoses from a district hospital. Prosthetics and Orthotics Int., 1989, 13, 29-35. 7 Lord, M. and Jones, D. Issues and themes in computer aided design for external prosthetics and orthotics. J . Biomed. Enyng, 1988,10,491498. 8 Smith, P. J. Shoe design and manufacture by computer. Con,ference on Design engineering, Birmingham, 1984 (Institution of Mechanical Engineers). 9 Newman, W. M. and Spronll, R. F. Principles of inteructiw computer graphics, 1981 (McGraw-Hill). 10 Lord, M. and Foulston, J. Clinical trial of a computer-aided design system for oathopaedic shoe upper design. Prosthetics and Orthotics Int., 1991, 15, 11-17. 11 Lord, M. and Travis, R. P. Surface modelling the foot from OSIRIS scans. In Surface topography and body deformity (Eds H. Neugebauer and G. Windischbauer), 1990 (Gustav Fischer Verlag). 12 Duncan, J. P. and Mair, S. G. Sculptured surfaces in engineering and medicine, 1983 (Cambridge University Press). 13 Saunders, C. G. and Vickers, G. W. A generalised approach to the replication of cylindrical bodies with compound curvature. Trans. A S M E , 1984,106,7&76.
Proc Instn Mech Engrs Vol 205
@ IMechE 1991 Downloaded from pih.sagepub.com at UQ Library on July 15, 2015
