A Quantitative Method For Designing Manual Socket Modifications

Introduction

CAD/CAM (Computer Aided Design / Computer Aided Manufacture) has become a viable clinical tool for the field of prosthetics and orthotics. The flexibility, speed, precision, and efficiency of these systems have made them useful for a wide range of applications; such as, prosthetic socket design, custom wheelchair seating, spinal orthoses, and foot orthoses. Generally, clinical prosthetists use CAD/CAM to produce a positive model of a residual limb from which a prosthetic socket can be manufactured. The CAD/CAM process provides a controlled method for shape modification, a more accurate method for positive mould fabrication, a decrease in production time, and a more efficient platform from which to service remote areas (i.e., mobile clinic).

Prosthetic CAD/CAM systems were the first to emerge in a clinical setting; possibly because it was easy and cost effective to digitize the conical shape of a trans-tibial amputee's residual limb and carve the modified positive. Improvements in CAD software let clinicians make almost any residual limb shape modification on-screen. A prosthetist uses computer modification tools to outline a modification region, specify the points of maximum change, and set the amount of modification. While these tools are effective, the prosthetist must be able to visualize socket modifications on a two dimensional (2D) screen as opposed to hand-sculpted modifications on a physical object. Experience with clinical CAD/CAM applications has shown that the transfer of manual prosthetic modification skills to a computer system is not an easy or time-efficient task. This knowledge-transfer problem is expanded when the individuality of clinical modification procedures are considered (i.e., each prosthetist has a slightly, or very, different modification approach). In most cases, a prosthetist will learn to modify a shape on a CAD system by trial and error.

One approach to easing the manual-to-CAD/CAM transition is to develop a method of quantitatively defining manual socket modifications. Once a manual modification technique has been quantitatively defined, the numerical modifications can be transferred to a CAD system as a template or overlay. A template will allow the user to apply their manual modification technique to a shape in one step. Custom modifications can then be made to the averaged modification pattern. This quantitative method should improve the efficiency and effectiveness of moving from traditional to computer socket design.

This document will describe a project for quantitatively defining manual socket modifications. The quantitative method will be used to produce a template for CAD/CAM socket fabrication. Validation of the template will involve comparing CAD/CAM produced sockets against manually manufactured sockets.

Methods

Subjects

All subjects involved with this study were recruited through the Prosthetics and Orthotics Service at The Rehabilitation Centre (Ottawa, Canada). These trans-tibial (TT) amputees used their PTB prosthesis with supracondylar suspension as their main ambulatory assistive device. Four prosthetists were involved with this study; however, only one prosthetist was involved in the validation phase (fitting patients and evaluating socket fit). The other prosthetists provided clinical input and helped test the CADVIEW program. The standard modification pattern procedure was based on clinical feedback from all prosthetists. Seven subjects with TT amputations were recruited to define the standard modification pattern. An additional four experienced prosthetic wearers were recruited for a pretest of the CAD/CAM rectification pattern. A total of 13 subjects with TT amputations, who were more than one year postoperative, were recruited for the validation phase of this project.

Equipment

All CAD/CAM software and hardware were available in the Prosthetics and Orthotics Service of the Rehabilitation Centre (Ottawa, Canada). The average modification pattern shapes were digitized with the CANFIT-PLUS CAD system and the Shapemaker software package was used to produce all test sockets. The CANFIT-PLUS system ran on a 486-33 MHz computer and Shapemaker ran on an IBM Pentium-166 Mhz computer. The gait testing facilities at The Rehabilitation Centre include one AMTI force platform, an Ariel Performance Analysis System (video kinematic/kinetic analysis), electrogoniometers, and a proprietary EMG data collection system.

To develop a quantitative approach for defining prosthetist-specific socket modification patterns, a software program was written to compare original and modified residual limb shapes. By entering a series of pre and post-modification socket shapes into the CADVIEW software, common modification areas could be averaged to produce a generalized rectification pattern suitable for use with a CAD/CAM system.

Digitizing Protocol

Standard cast and modification protocols were developed to ensure consistent application of the modification template and consistent manual techniques (i.e., changes between subjects were due to individual differences and not differences in casting and modification technique). The digitizing protocol steps for defining manual modification included:

Comparison Process

The comparison process involved the following steps:

pic

Validation

Before formally validating the CAD/CAM technique, a pretest was performed involving four experienced users of PTB sockets with supracondylar suspension. After each subject was fit with a CAD/CAM produced socket, the prosthetist and the end-user assessed the success of the standardized modification pattern. The success was based on clinical criterion and whether major modifications were required.

Each of the 13 validation subjects were fit with a CAD/CAM produced socket and, if their current socket was unsatisfactory, fit with a new conventional socket. The clients wore their new prosthesis for at least two weeks before completing a questionnaire and having their gait evaluated.

A clinician questionnaire and a client questionnaire were used to assess satisfaction with the conventional and CAD/CAM produced socket. The clinician questionnaire recorded information on the prescribed device, the time required to fit the client, clinician satisfaction with the manufactured socket, the number and type of modifications required for final fitting, and a qualitative assessment of their walking gait. One questionnaire was completed for each socket at the time of testing. The same components were used for both prostheses. The client questionnaire inquired about comfort, security, ease of gait, pain and pressure problems (fit), general satisfaction, and general comments.

Quantitative gait analysis was used to ensure that no significant walking pattern differences were produced by wearing either a CAD/CAM or a conventional socket. All gait testing took place in the Gait and Motion Analysis Laboratory at The Rehabilitation Centre. The gait analysis equipment in this study included an AMTI force platform, the Ariel Performance Analysis System (APAS), an IBM compatible 486 computer, and a video camera. The AMTI /APAS configuration was used to sample two seconds of ground reaction force data for each trial at 200 Hz.

After joint markers were placed on the subject, the subject walked at a natural cadence along a 10 metre walkway until they felt comfortable in the laboratory and consistently stepped on the force platform. The subject walked in the same direction to collect 12 good trials. Following each data collection session, the force data were digitally filtered at 12 Hz and transferred to the data processing computer. The APAS system was used to capture a digital video clip for each trial and digitize the 2D marker positions. After the data had been transformed, most marker data were digitally filtered at 10 Hz. In a few instances, a six or eight hertz filter setting was required (these trials had poor video quality and, therefore, more digitizing error at the foot or knee markers). The APAS graphing utility was used to obtain the stride length, stride time, and walking speed by subtracting toe marker positions and times at successive toe-off events. Stance time was calculated from the ground reaction force data.

Force post-processing involved importing a subject's filtered data into Quattro Pro so that all 12 trials could be averaged. Each trial was normalized to 100 percent of stride using linear interpolation. The average and standard deviations were calculated at one percent intervals. Peak forces were calculated for each trial from the filtered data (i.e., data not normalized and not averaged). These peaks included the maximum lateral-horizontal force (Fx), the minimum breaking force (Fy-brake), the maximum push-off force (Fy-push), and the maximum vertical forces (Fz-brake, Fz-push) at weight acceptance and push-off.

A separate, Microsoft Windows program was written to read the filtered APAS force data and calculate impulses. This program displayed the Fx, Fy, and Fz force-time curves for each trial and allowed a user to graphically remove a bias or examine individual data values. All negative and positive impulse values for the currently displayed curve were automatically calculated by the software. Impulse values for Fx, Fy-brake, Fy-push, Fz-brake, and Fz-push were copied into Quattro Pro for statistical analysis.

Results

Pre-test

While each subject's socket required specific modifications, some common changes were required for all four pre-test subjects. To accommodate individual variations in anatomical structure, modification locations were changed for each subject. For the same reason, apex point positions were also changed for some modifications. The size of certain modifications also had to be changed due to Shapemaker's inability to adequately scale the template for long or short shapes. Creation of a long modification outline, while maintaining the same modification shapes, alleviated some of these template application problems.

Table 1: Characteristics of pre-test subjects

# Age (yrs) Ht (m) Wt (kg) Amp. Site Limb Limb Condition
1 74 1.82 91 Right TT (traumatic) Short, boney Good, verrucose hyperplasia
2 63 1.7 91 Right TT (vascular) Short Good
3 62 1.72 84 Right TT (traumatic) Short, severe scar tissue,graft Skin condition under control
4 49 1.8 86 Left TT (traumatic) Long, good stump Good

Following pre-evaluation, the medial tibial flare modification was split into two sections. This created the appropriate medial tibial flare relief when applying a Shapemaker template. A long socket template was also created and used in cases where limb lengths were beyond the capabilities of Shapemaker to adequately scale the template. The revised modification template was considered appropriate for clinical use and full validation testing.

Questionnaire Results

A client questionnaire was used to obtain each subject's perspective on comfort and function for the manual and CAD/CAM prostheses. Wilcoxen signed ranks test results showed no significant differences (p<0.05) between the two prostheses based on comfort, ability to walk, and overall satisfaction. McNemar test statistics also showed no significant between-group differences (p<0.05) on the basis of pain (p=0.69) and perceived safety (p=1.0) during prosthetic use. These results supported the premise that the new CAD/CAM design technique can produce a socket that is considered as good as a manually produced socket.

Examination of individual subject data showed that, in four cases, the CAD/CAM prosthesis was considered superior to the manual prosthesis. In each case, the ratings were only one level higher. Three other subjects considered their manual prosthesis superior. These subject were long term prosthetic users who were very satisfied with their current prosthesis. Two of these subjects experienced some discomfort when ambulating with their CAD/CAM prosthesis. The discomfort areas were the fibular head region and the medial patella. All of the CAD/CAM sockets were considered safe to use; however, one subject did not consider their original prosthesis safe. The subject was not able to explain this result.

The prosthetist questionnaire results also supported the premise that the CAD/CAM technique can produce a socket of equal quality as a manually produced socket. No significant differences (p<0.05) were found between the two groups for walking gait and socket fit (Wilcoxen signed ranks test at p<0.05).

The prosthetist graded CAD/CAM socket fit as superior in four cases and manual socket fit as superior in one case. These results compared well with the results from the client questionnaire; however, the prosthetist and client differed in opinion in two instances. In both these cases, the client liked the manual socket better while the prosthetist rated both sockets the same. One case that differed was for a long term prosthetic user who did not like a hard socket. For the other case, the subject experienced some medial patellar discomfort when using the CAD/CAM prosthesis. This discomfort was resolved after the second gait analysis session. The prosthetist considered the quality of walking gait to be the same for both groups in all but one case. All sockets were considered safe.

On average, 1.5 (s=0.63) CAD/CAM sockets were required to produce an acceptable device. Eight out of thirteen sockets were acceptable on the first attempt and one socket required three attempts to obtain a satisfactory result. Three attempts were necessary for the one unsuccessful trial (i.e., the trial where the client and the prosthetist considered the manual socket to be superior). The prosthetist made small modifications with a heat gun in 62% of the socket fittings.

Gait Results

Gait analysis results from the manual and CAD/CAM socket groups were very similar. T-test analysis results showed no significant differences between groups (p<0.05) in all cases except peak vertical forces on the amputated side. Data from both groups were significantly correlated (p<0.05) for all measures. Gait results from the amputated limb were collected for all subjects. The y-ratio force measure was calculated by dividing the peak braking force by the peak push-off force for each trial. The braking impulse was divided by the push-off impulse to calculate the impulse y-ratio.

Gait data from the non-amputated limb were collected for nine subjects. No significant differences (p<0.05) were found for any measures. All measures on the non-amputated side were significantly correlated (p<0.05).

The force plate data were ensemble averaged for each subject. Root mean square error (RMSE) and Pearson correlation coefficients were calculated from the manual socket and CAD/CAM socket ensemble curves. Since all but one correlation coefficient were greater than 0.93, the CAD/CAM prostheses did not affect the force/time curve shapes. For most subjects, the RMSE values were low. This suggested that the CAD/CAM prosthesis did not affect the force/time curve magnitudes. All cases where the RMSE was more than five percent of the full range were from the subjects with the most variable gait. The Fx force component was the most variable within-subject measure and had the largest relative RMSE values.

Discussion

The reliance on hand-sculpting techniques in the field of prosthetics has contributed to the development of prosthetist-specific methods for socket design. Since the optimal method for designing a functional and comfortable socket has yet to be discovered, the client must rely on an individual prosthetist's clinical judgement to make and fit a TT prosthesis.

Though a prosthetist can modify a socket successfully, many are unable to define exactly what was done to the positive model. During modification, the original shape is lost as material is added and removed. Because modifications are made over the entire shape and not as a series of individual changes, picking out exactly how the residual limb shape was modified is difficult. The inability to define how individual prosthetists modify a socket can impede the transition from manual techniques to CAD/CAM. To address this issue, a Microsoft Windows software program was written to display, analyse, and compare manual and CAD/CAM socket shapes. This tools was used to reverse-engineer a prosthetist's manual modification technique and generate a CAD template.

Validation was an important part of this initiative to ensure that the new method for defining CAD/CAM modifications produced a socket that was of equal, or better, quality than a socket that was designed by traditional techniques. To complete this phase, client questionnaires, prosthetist questionnaires, and gait analyses were used to assess each socket.

A client questionnaire was used to obtain the client's perspective on comfort, function, safety, and satisfaction with their prosthesis. The subject's opinion on socket fit and function is important since much of the prosthetic fitting process depends on subjective feedback from the client. No significant differences were found between the CAD/CAM and manual modification groups for any of the questionnaire responses. Since the prosthetist questionnaire produced similar results, the opinions rendered in these questionnaires can be considered valid. These results indicate that the prosthetist and the subjects considered sockets designed using the CAD/CAM technique to be at least as good as the manually designed sockets.

Examination of the questionnaire data from individual cases provided insight into the clinical realities of using a CAD modification template. Of the thirteen test cases, one socket fitting can definitely be considered unsuccessful. Upon discussion with this subject's regular prosthetist, it can take months of trial and error to successfully fit this person with a prosthesis (using manual methods). This difficulty is likely a combination of the subject's boney/high contour residual limb shape, scar tissue, and unorthodox walking gait. This person is also very stoic and will put up with some discomfort before asking for an adjustment. Even though the CAD/CAM socket was not as good as the subject's usual device, the level of success was considered typical for this client. Although this does help to explain the results, it does not change the fact that the socket fitting was unsuccessful.

In two other cases, fitting success was not as clear. These two subject preferred the manual socket over the CAD/CAM socket; however, the prosthetist considered the CAD/CAM sockets to be as good as the manual units. In one case, the subject reported some discomfort in the medial patellar region. Unfortunately, the subject did not report this discomfort to the prosthetist until returning for post-evaluation. The prosthetist could have easily corrected this problem with a heat-gun modification. Since this discomfort was not present when ambulating with his regular prosthesis, it is understandable that the subject would assign higher ratings to his manually designed socket. The third subject indicated that the hard CAD socket was too different from his old, Pelite lined, socket. To correct this, a second socket was made that incorporated a soft liner. The subject felt more comfortable with the soft liner; however, he was unable to explain why he still preferred the manual socket. It was interesting to note that this subject was the only person that did not return for adjustments after the initial fitting. This may have meant that the socket did not require adjustments or that the subject did not want to make an effort to have an optimally fitting CAD socket (i.e., just get the socket and finish his commitment as a subject in this project).

In four cases, the client considered the CAD/CAM socket superior. The prosthetist concurred on three of these cases - he considered the fourth case to be as good a fit as the previous socket. This consensus between the subject and the prosthetist adds credibility to the idea that the new CAD/CAM modification technique is capable of creating a prosthetic socket that is better than a person's current device. Although it was clear that some subjects preferred the CAD/CAM socket, the reasons for this preference were diverse.

In one case, the subject was experiencing some pain when walking with the old prosthesis. This pain was not present when the subject was retested with the new prosthesis. The resolution of this pain may have been related to the new socket or the pain may have resolved itself over the two-week inter-test interval. For the second and third cases, the new socket required three to four ply less socks than the manually produced socket. The reduction in socket volume may have lead to more prosthetic control during gait. A tighter socket may also have felt more comfortable since it would have had to conform better to the subject's anatomy. It was difficult to identify one factor that could describe why the last subject preferred the CAD/CAM socket. This subject was extremely satisfied with his new socket and requested that a permanent prosthesis be fabricated with the same design. Since the fitting session was extremely easy it may be concluded that the prosthetist made the correct choices to produce an optimal socket for this client. Some clients also stated that they liked the foam pad that was positioned at the distal end of the CAD/CAM prosthesis; however, this remark did not necessarily correspond with superior rating for either prosthetic socket style.

Since the CAD/CAM prosthesis was compared with the subject's current prosthesis, it was not possible to blind the subject or the prosthetist as to what device was being tested. The novelty of using a new socket may also have contributed to a superior rating; however, bias towards a new device may also have contributed to the inferior ratings.

Both manual and CAD/CAM methods sometime require that more than one socket be fabricated before a successful fit is achieved. For this study, the average of 1.5 iterations was comparable with CAD/CAM results in the literature and falls within the expected clinical range. It was not surprising that the subject who was not successfully fit required three iterations before an acceptable socket was designed. There were no visible trends between satisfaction with one type of prosthesis and the number of iterations that were required to obtain an acceptable result.

The gait analysis results supported the hypothesis that there was no difference between the CAD/CAM socket group and the manually produced socket group. In almost all cases, there were high correlations and small between-mean differences. These results were consistent for discrete measures and for ensemble averaged curve comparisons.

The stride parameter results from this study were comparable with similar results in the literature. These results were also very similar when comparing the two groups. Since the stride length, stride time, and walking speed results were so close, between group gait comparisons should not be substantially affected by variations in walking speed.

Between group, ground reaction force comparisons produced the only significantly different measures. On the amputated side, the vertical ground reaction forces that were produced during the manual socket trials were significantly higher than vertical forces from the CAD/CAM socket trials. The average vertical peak forces were also lower on the non-amputated side; however, these results were not significant. Since the differences in vertical impulse values were small, it can be concluded that the vertical forces on weight acceptance and push-off were redistributed over each of these phases. Examination of the averaged force/time curves for each subject supported this idea since curves with lower peak forces compensated by having a lower slope, and hence a more equal area. Other methods for achieving similar vertical impulses included a reduced unweighting phase and a more abrupt push-off (thereby increasing the area under the force-time curve).

The medial-lateral component of the horizontal force was the most variable measure. This is not an uncommon finding when testing people with, or without, a lower extremity amputation. Even with the high variability, each curve had the same general shape. There was no clinically identifiable between-group differences for the medial-lateral ground reaction force curves.

It was interesting to note that people who preferred the CAD/CAM socket had the largest reduction in peak vertical ground reaction forces. No such trend was apparent for the people who preferred their manually designed socket. The people who liked the CAD/CAM prosthesis also had lower Fx impulse values, higher Fy braking impulses, and larger push-off impulse values. While these results were not significant, they may help explain the success of these new prostheses. Lower Fx impulses may have indicated that there was less total body centre of gravity movement away from the midline. This could improve the subject's perception of balance. The higher breaking and push-off impulse values could indicate that these subjects were making better use of their prosthesis to stop their forward progression and for propulsion. Improved force transfer from the prosthesis to the ground should result in overall improvements in walking gait and, as a result, in improved client satisfaction. No observable trends were present for the people who preferred their manually produced prosthesis.

The averaged force/time curve shapes were similar in almost all cases (i.e., when comparing CAD/CAM versus manual trials for the same subject). Some of the differences that were observed by examining the ensemble averaged data included the following:

Even with these documented variations, the force/time curves from the CAD/CAM trials were usually within one standard deviation of similar data from the manual trials.

Except for the differences in peak vertical forces, there were no clinically or statistically relevant differences between the manual socket and CAD/CAM socket groups. This result supports using the modification outline generating process from this thesis to develop a clinically viable CAD/CAM template. The modification outline generating process may also have helped the prosthetist make the transition from manual socket design to computer-aided design.