Central fabrication: carved positive assessment

Original Research Report Prosthetics and Orthotics International 35(1) 81–89 ! The International Society for Prosthetics and Orthotics 2011 Reprints and permissions: sagepub.co.uk/journalsPermissions.nav DOI: 10.1177/0309364610394476 poi.sagepub.com Central fabrication: carved positive assessment Joan E Sanders, Michael R Severance, Timothy R Myers and Marcia A Ciol Abstract Background: It is estimated that only 24% of practitioners use CAD/CAM regularly. Socket manufacturing error may be a source of the limited use of central fabrication. Objectives: The purpose of this study was to investigate the differences in shape between computer-manufactured, centrally fabricated carved models and electronic file shapes, to determine if carving was a major source of socket manufacturing error in central fabrication. Study design: Experimental, mechanical assessment. Methods: Three different trans-tibial model shapes were sent electronically to each of 10 central fabrication facilities for the fabrication of positive foam models. A custom mechanical digitizer and alignment algorithm were used to measure the model shapes and then compare them with the electronic file shapes. Results: Volume differences between the models and the electronic file shapes ranged from 4.2% to 1.0%, and averaged 0.9 (SD ¼ 1.1)%. Mean radial error ranged from 1.2 mm to 0.3 mm and averaged 0.3 (SD ¼ 0.3) mm. Inter-quartile range was between 0.3 mm and 2.7 mm and averaged 0.6 (SD ¼ 0.5) mm. The models were significantly smaller than sockets made from the same electronic file shapes (p < 0.01), but the range of mean radial error and the interquartile range were not significantly different between the models and sockets. Conclusions: The results demonstrated that there was considerable variability in model quality among central fabricators in the industry, and that carving was not the sole source of socket fabrication error. Clinical relevance The results provide insight into the severity and nature of carving error by central fabrication facilities. Because we found a wide range of model quality, there is not a consistent fabrication problem across the industry, but instead some central fabrication facilities practice the art of model fabrication better than others. Keywords CAD/CAM, fabrication techniques, prosthetic design, rehabilitation of amputees Date received: 19 November 2010; accepted: 24 November 2010 Introduction Central fabrication facilities fulfil an important role in trans-tibial amputee prosthetic care by providing custom remote manufacturing services for clinical prosthetists. Upon receiving an electronic file of the desired socket shape from the customer, central fabricators carve a solid positive model, then thermoform a polymer sheet or cone over it (Figure 1). After removing the model material from the inside of the socket and trimming the brim, they send the socket to the customer for fitting to the amputee patient. When performed properly, this process is fast, efficient and reduces capital equipment expenses for the practitioner compared to performing fabrication in-house. A May 2009 publication1 identified 42 facilities in the US offering central fabrication of trans-tibial prosthetic sockets. Assessments have been reported in the scientific literature for different stages of the CAD/CAM process. Geil2 demonstrated that for capturing the limb shape, University of Washington, Seattle, USA. Corresponding author: Joan E Sanders, University of Washington, PO Box 355061, 3720 15th Ave NE, Seattle, WA 98195, USA Email: jsanders@u.washington.edu Downloaded from poi.sagepub.com by guest on July 17, 2016 82 Prosthetics and Orthotics International 35(1) manual, optical and electromagnetic shape-capture systems produced anthropometric measurements similar to each other. McGarry et al.3 found that an electromagnetic contact shape-sensing system did not perform as well on trans-tibial models as on cylindrical models. The number of CAD/CAM sockets required to properly fit amputee subjects ranged from 1.5 to 5.0, depending on the investigation.4,5,6,7,8,9 Several authors have reported that less time is required to achieve an acceptable fit using computer manufacturing versus traditional hand methods,4–5,6,7,10 and one author claimed that cost was reduced.11 A practitioner using central fabrication does not need to pay for lab space for fabrication, technician costs to make sockets, fabrication equipment, supply and maintenance costs to make positive models and form sockets, and insurance premiums to cover these expenses. However, current reimbursement practices allow this practitioner to charge and receive the same amount per patient as a practitioner who does incur these costs. Those who use it effectively claim that the cost savings with CAD/CAM will drive traditional methods out of the market.12 There is agreement that computer manufacturing is the way of the future in prosthetics.4,6,11,13,14,15,16,17,18 The purpose of this research was to assess whether the first of the two manufacturing stages of central fabrication, the carving phase (Figure 1), was a major source of fabrication error using CAD/CAM. The shapes of carved positive models were compared to the electronic file shapes used to create them. We also compared the results with centrally-fabricated check sockets to gain insight into how carving errors compared with socket shape errors. Methods Socket designs Three model designs were tested. The three shapes were example amputee patient cases from the Capture limb shape Design socket shape Carve positive model Form socket Fit socket to patient ShapeMakerTM (Sheck and Siress, Chicago, Illinois) prosthetic design software package. They were the same shapes as used in a socket shape assessment study reported previously.19 Socket A was 12.8 cm in length from the mid-patellar tendon to the distal end, of a traditional patellar tendon-bearing design, and included 68 slices. It was a short, conical shape for a bony residual limb with a prominent fibular head. Socket B was 14.2 cm in length, of a total surfacebearing design, and included 50 slices. It was for a fleshy, cylindrical, bulbous residual limb with minimal bony prominences. Socket C was 15.7 cm in length, cylindrical in shape with prominent bony landmarks, and included 48 slices. It was for a patient with a bony residual limb, similar to Socket A, but more cylindrical in shape. The socket shape data were stored in an American Academy of Orthotists and Prosthetists (AAOP) file format at 72 points/slice. Central fabrication facilities Ten central fabrication facilities were contracted to manufacture positive models of each of the three socket shapes, using shape files sent to them via email. The cost per model ranged from US$47 to $75 in 2007, not including tax or shipping. All models were made of a polyurethane-based foam. One model was ordered at a time, as opposed to ordering all three models together. According to company representatives, none of the companies modified the shape files before carving, though all except companies 2 and 5 stated that they lightly sanded the models before forming sockets in their facilities. All added a nylon sock or layer of silicone when forming check sockets over the models. Companies may have used different commercially available carvers to create their models and may have differed in their calibration practices, bit sizes, computer-aided manufacturing software and other manufacturing parameters. Because the purpose of this research was to evaluate centrally fabricated models as opposed to other variables, we did not control these variables in this study. Check sockets from a previous study19 were digitized in this research at the same resolution as used for the models, so that comparisons between the sockets and models could be made. The same manufacturers were used in both studies and the same reference number used to identify each company, except for company 7 which was a different manufacturer. Model digitization Figure 1. Stages of computer socket manufacturing. Central fabrication facilities provide services for the carving and forming stages of the process. A custom mechanical digitizer was used to measure model shapes (Figure 2). We used a custom instrument because commercial digitizers and scanners lacked the Downloaded from poi.sagepub.com by guest on July 17, 2016 83 Sanders et al. Table 1. Accuracies of the measures of interest. Feature Volume difference 0.1% Mean radii difference 0.05 mm Histograms/interquartile range 0.08 mm Shape difference images 0.08 mm (4% of the color scale) RVDT Stylus arm Model Accuracy Linear slide rail Gearbox Base coordinate system (r-z), at each angular interval. The stylus ball’s centres were then projected outward along the spline normals (in r-z) by the distance of the stylus ball radius. Splines were then fitted to the projected data. For all positive models tested, the digitizer was set at a slice spacing of 0.8 mm and a density of 800 points per slice. Thus the resolution of the digitizer data was far greater than that of the electronic file shapes, a necessary condition to be able to accurately align and compare the two shapes. It took about 6 h to digitize each model. To compare the results of the present investigation of positive models with those of the check sockets from a previous investigation,19 the check sockets were re-digitized using these digitizer settings, after removing the custom gearbox that offset the motor axis and recalibrating. Shape analysis Figure 2. Instrument used to measure model shape. The model was rotated using a motor in the base while stylus arm position was measured with a rotational variable differential transformer (RVDT) at the top of the arm and a linear displacement transducer (LDT) within the linear slide rail. The gearbox offset the motor axis so that the stylus arm was not excessively angulated during operation. performance qualities necessary to conduct this analysis. A full contact method was used for assessment, similar to that described previously.19 A spring-loaded arm contacted the positive model surface such that the tip of the stylus was very accurately measured using position sensors on the stylus arm (RVDT, R30A with an ATA 2001, Schaevitz, Hampton, Virginia) and the slide rail (LDT, BTL-5-A/C/E/G1-M457-R-S32, Balluff, Florence, Kentucky). A custom gearbox offset the motor axis to prevent excessive angulation of the stylus arm during operation. The offset distance was 88.9 mm and the gear ratio was 1 : 1. A detailed evaluation of the digitizer was conducted and is reported elsewhere.20 Accuracies for the measurements of interest are shown in Table 1. Data were collected at equally spaced angular intervals of 0.45 degrees. Contact between the model and the stylus ball was not necessarily along the radial axis in the cross-section of the slice, due to the varying curvature of the model surface. Therefore, we recorded positions of the centre of the stylus ball. Splines were fitted to these data in radial-longitudinal planes in a cylindrical To align the measured model shapes to the electronic file shapes, an optimization procedure that minimized shape difference while maximizing shape similarity was used. The algorithm has been described in detail elsewhere,21 and its application to prosthetic socket assessment discussed.19 The shape alignment algorithm is an optimization procedure that uses a weighted linear combination of maximized global similarity and maximized local shape similarity.21 To maximize global similarity, the minimum weight-matching variant of the bottleneckmatching method is used, minimizing the sum of the absolute differences between the two corresponding radii of the two point sets.22 Clinically, this process corresponds to minimizing the volume difference between two shapes. Then another algorithm is implemented to maximize local shape similarity by summing the dot products of corresponding surface normals. This twostep algorithm has proven extremely robust in prosthetics applications. It has been used to effectively characterize residual limb shape changes measured over time,23 and shape differences between sockets and electronic shape files.19 It represents an advancement over previous shape-alignment efforts that used exclusively a minimization of volume difference,24 anatomical landmarks25–26 or top and bottom slice centroids.27 The optimization criteria were a minimization of mean error and alignment of surface normals, weighted in a ratio of 0.8 to 0.2. This weighting ratio was Downloaded from poi.sagepub.com by guest on July 17, 2016 84 established from experience in prior investigations of sockets.19 All points of the electronic file shape (4896, 3600 and 3456 for models A, B and C, respectively) were used in the analysis. All 10 models were aligned with the reference shape, i.e. the electronic file shape, within the same optimization procedure to ensure that the same length of model was analysed. The model length selected was the maximum length common to all eleven shapes. The brims of the model shapes and socket shapes were trimmed on the computer using standard trimline locations for each of the three designs. Only the surfaces of the models within the trimlines were considered in the volume, radii and shape analyses described below. The area in each cross-section was calculated using the ‘polyarea’ function in Matlab (Mathworks, Natick, Massachusetts), then multiplied by the distance between sample cross-sections to determine sectional volume. The sectional volumes were summed to determine the total volume. Volume differences and radial error differences between the model shapes and electronic file shapes, as well as between the socket shapes and electronic file shapes, were calculated. Histograms of the radial error were created at an increment of 0.1 mm, a value slightly greater than the error in the digitizer measurement. The interquartile range (IQR) was calculated as the range in millimetres over which the middle 50% of the data spanned the median radial error. Results Model-AAOP (% volume difference) Prosthetics and Orthotics International 35(1) 2 1 0 –1 A B C –2 –3 –4 –5 1 2 3 4 5 6 7 Company # 8 9 10 Figure 3. Volume differences between models and electronic file shapes. Results are expressed as percentage difference relative to the electronic file shapes. For all 10 companies, anterior and posterior brim regions were typically oversized, while the region between the patellar tendon bar and the tibial tuberosity was undersized. We investigated how the mismatches at these two locations affected the measured total volume differences between the carved models and the electronic file shapes. The brim and patellar tendon region on the anterior aspect and the brim region on the posterior aspect for each model were isolated. Results showed that the volume contributions for the anterior region to the volume difference between the models and electronic file shapes for the 30 models averaged 0.02 (SD ¼ 0.10)% and for the posterior region averaged 0.11 (SD ¼ 0.09)%. Volume differences Volume differences between the carved models and the electronic file shapes ranged from 4.2% to 1.0% (Figure 3), and averaged 0.9 (SD ¼ 1.1)%. Five of the carved models were larger than the electronic file shapes, while 25 were smaller. A one-way analysis of variance showed that there was no statistically significant difference between the mean volume differences from model shapes A, B and C (p ¼ 0.29). The range in the volume differences for the three models from a company, i.e. maximum volume difference (model-electronic file shape)-minimum volume difference (model-electronic file shape), averaged 1.2 (SD ¼ 1.0)% and ranged from 0.3% to 3.3%. The 1.2% range corresponded to a radial distance of approximately 0.4 mm. At least one model from each of companies 5, 8 and 9 had a much larger percent volume difference than the other models (Figure 3). Thus their ranges in volume difference (3.3%, 1.7% and 2.5%, respectively) were much greater than for the other seven companies (all <0.9%). These three companies were thus less consistent in their performance than the other seven. Distribution of radii differences Mean radii differences between the carved models and electronic file shapes averaged 0.3 (SD ¼ 0.3) mm for the 30 models. However, differences were considerably more negative, between 1.2 mm and 0.6 mm, for five of the models (two each from companies 8 and 9, and one from company 5) than the other 25 models. Thus mean radial errors and their ranges appeared more negative for companies 5, 8, and 9 than the other companies (Figure 4a). For companies 1–4, 6, 7 and 10, histograms of the radii differences were relatively consistent in shape for the three models from the same company. Histograms for companies 2, 6 and 10 were relatively sharp and narrow (Figure 5a), and those for companies 1, 3, 4 and 7 were shorter and wider (Figure 5b). The IQR averaged 0.6 (SD ¼ 0.5) mm for all models and ranged from 0.3 to 2.7 mm. However, the IQR was considerably less, ranging from 0.3 to 0.5 mm for models from companies 2, 6 and 10. It was between 0.3 and 0.8 mm for models from seven of the companies, namely 1–4, 6, Downloaded from poi.sagepub.com by guest on July 17, 2016 85 Sanders et al. (a) 1.0 Model A Model B Model C Mean Radial Error (mm) 0.5 0.0 –0.5 –1.0 –1.5 –2.0 0 1 2 3 4 5 6 7 Company # 8 9 10 3 4 5 6 7 Company # 8 9 10 (b) 2.0 Socket A Socket B Socket C Mean Radial Error (mm) 1.5 1.0 0.5 0.0 –0.5 –1.0 0 1 2 Figure 4. Mean radii differences compared with electronic file shapes for (a) models and (b) sockets. Vertical axis is 3.0 mm for both plots. 7 and 10. Histograms for companies 5, 8 and 9 were inconsistent, with at least one model from each company showing a flattened or distorted shape (Figure 5c) and an IQR  0.9 mm. Check sockets versus models The number of check sockets of the 30 tested within a 1.0% volume error was comparable to that for the models. Eighteen of the 30 sockets and 20 of the 30 models were within a 1.0% volume of the electronic file shape. For companies common to both investigations (n ¼ 9), a paired t-test showed that the models were smaller than the sockets (p < 0.01). The mean radial error for models was 0.3 (SD ¼ 0.3) mm while that for sockets was 0.1 (SD ¼ 0.5) mm (Figure 4). Using data from companies common to both investigations (n ¼ 9), a paired t-test showed that there was no significant difference in the range of mean radial error between the models and sockets (p ¼ 0.48). The mean range for the nine companies was 0.3 (SD ¼ 0.3) mm for the models, and 0.4 (SD ¼ 0.4) mm for the sockets. Another paired t-test showed that there was no significant difference in IQR between the models and sockets (p ¼ 0.64) for the nine companies common to both investigations. The IQR averaged 0.6 (SD ¼ 0.5) mm for the models and 0.6 (SD ¼ 0.3) mm for the sockets. To visualize results from the model analysis simultaneously with those from the socket analysis, we created a plot of mean radial error for models versus mean radial error for sockets for companies common to both investigations (n ¼ 9) (Figure 6). The box for each company is the smallest possible size such that it contains all three model/socket pairs tested. The red line is the zero axis for model mean radial error, and the blue line is the zero axis for socket mean radial error. Thus companies with smaller boxes that are nearer the blue line demonstrate better and more consistent carving and forming results than companies with larger boxes or boxes further from the blue line. Companies 1, 2 and 3 show the smallest box sizes (greatest consistency) and are nearest to the blue line (lowest socket error), and thus demonstrate the best and most consistent carving and forming results. Companies 4, 6 and 10 have boxes of low vertical dimension, indicating that their carving practices were relatively consistent. However, the large horizontal dimensions of their boxes and their further distance from the blue line compared with companies 1, 2 and 3 indicate that their forming practices are not as good. Companies 5 and 9 have large vertical dimension boxes, indicating inconsistent carving. Company 8 is inconsistent in both carving and forming, demonstrating the largest box of all companies tested. Discussion The potential benefits of CAD/CAM and central fabrication have yet to be realized by much of the clinical prosthetics community. It is estimated that only 24% of practitioners use CAD/CAM regularly.28 Because socket manufacturing error might be a source of the limited use of central fabrication,19 the purpose of this investigation was to characterize manufacturing error during the initial stage of central fabrication, the carving stage, to determine if carving was a major source of error. Volume and mean radii differences We considered why most of the models, 25 of 30, were undersized compared with the electronic file shapes. The anticipation of adding a nylon sheath or silicone Downloaded from poi.sagepub.com by guest on July 17, 2016 86 Prosthetics and Orthotics International 35(1) (a) (b) IQR=0.4 mm (c) IQR=0.5 mm IQR=0.9 mm Figure 5. Histograms and shape difference maps. Histograms of radii differences between models and electronic file shapes (at 0.1 mm increments) are shown in the upper row. The vertical axis is the number of observations (points on the model surface). Results are shown for Model A for companies (a) 6, (b) 4 and (c) 5. Inter-quartile range (IQR) data are also shown. The corresponding shape difference maps are shown in the lower row. All scales are in mm. Figure 6. Mean radial error for models versus mean radial error for sockets. The box for each company is the smallest possible size such that it contains all three shapes tested. layer during subsequent socket forming, which all the manufacturers did, would seem a likely explanation. However, all manufacturers told us that they did not intentionally reduce the electronic file shapes to effect such compensation. Most manufacturers did state, however, that they sanded the carved models before sending them out to the customers or forming sockets over them. Sanding is commonly done to remove ridges on the model surface made by the carving bit, so that the model is smooth. All but two of the companies (companies 2 and 5) told us that they regularly sanded their models. Carved models from companies 2 and 5, Downloaded from poi.sagepub.com by guest on July 17, 2016 87 Sanders et al. however, did not show consistently larger models than the other companies (Figure 4a). This result suggests that companies individually tuned their calibration practices to meet socket shape needs rather than model needs. In other words, they calibrated around final socket shape rather than final model shape, since it was the socket shape that was of primary clinical interest. If sanding is indeed the main source of model undersizing then it is also a source of variability. It is done by hand, thus the amount of material removed and the area it is removed from depend on the person doing the sanding. Thus, despite being a computer-aided manufacturing process, model and socket forming is still sensitive to this human error. Distribution of radii differences For the same model shapes, some companies demonstrated lower IQR values than other companies (Figure 5a,b,c). This result indicates that there is not a consistent distortion limitation in the computer-aided model fabrication industry, but instead that some companies practise the art of model fabrication better than others. It is recognized that some companies might have different equipment or operate equipment at different settings, for example different carving speeds, and the results might simply reflect these variables. It is also possible that differences among companies reflect different fabrication practices, such as frequency of calibration or foam material used. The influence of these and other factors needs systematic investigation. A ranking of the contribution of each source of error to IQR would substantially benefit both CAD/CAM users and the manufacturing industry. Attention could then focus on overcoming or avoiding errors that cause the greatest distortion. Sockets versus models Because the sockets were not made from the same models as assessed in this study but from duplicates of them, quantitative conclusions cannot be drawn about how much carving error and how much forming error affected each company’s results. Only generalizations, as described below, are possible. As expected, most of the companies common to both investigations (six of nine) had greater ranges in volume error for the sockets than for the models (Figure 4a,b). The sockets represented error from both carving and forming, while the models represented error only from carving. Unless the errors cancelled each other out, carving and forming errors would be expected to accumulate. Inconsistency problems from other sources were present for some of the companies, particularly 5, 8 and 9, as shown in the data in Figure 4a. It remains a topic of future investigation to identify these sources of inconsistency. It is likewise relevant that IQRs were not significantly different between the collections of carved models and sockets for the companies common to both investigations. The shape distortions of the sockets were not consistently better or worse than the shapes of the models. Thus across the industry there is not a consistent loss of shape information during forming compared with carving. Clinical relevance of errors One of the reasons the nature of computer manufacturing error in making models and sockets is clinically relevant is that it dictates the practitioner’s options. Practitioners can more easily compensate for some errors than others. For example, most practitioners will learn to compensate for a consistent error. Consider a practitioner who receives carved models (or sockets) from a central fabrication facility that are always oversized by one sock, for example of 0.5 mm thickness. The practitioner will likely compensate and reduce his or her electronic file shapes by 0.5 mm before sending them to the facility. However, inconsistent volume error from model to model (or from socket to socket) is virtually impossible to accommodate and represents a most challenging clinical problem. The practitioner will be fitting patients with sockets of different quality, not because the practitioner is inconsistent, but because the fabrication is inconsistent. It is therefore important for practitioners to ask themselves, ‘Who made the error? Was it me or the central fabrication facility?’ If practitioners suspect a facility is inconsistent in their model or socket manufacturing, they should consider trying a different facility. If the inconsistency problem ceases then it is likely that the previous facility had a problem with inconsistent performance. Inconsistency in carving and in forming might have affected results from previous CAD/CAM studies where the number of sockets required to properly fit amputee patients was investigated. The mean number of sockets required, as reported in the literature,4–9 ranged from 1.5 to 5.0 sockets depending on the investigation. This high range may represent poor manufacturing consistency in some of the studies as opposed to differences in practitioners’ capabilities to use CAD/CAM effectively. It is important that future efforts delineate these two sources. The first is primarily an engineering manufacturing problem while the second is primarily a socket design problem. This investigation as well as our previous study on sockets19 provides insight into the magnitudes of error in central fabrication. But a question remains as to how Downloaded from poi.sagepub.com by guest on July 17, 2016 88 Percentage of models and sockets Prosthetics and Orthotics International 35(1) 70% 60% 50% Models Sockets 40% 30% 20% electronic file shapes, accounted for 10% of models and 17% of sockets. Thus about seven in eight of the models/ sockets tested here had percent volume differences that matched diurnal volume changes experienced by amputee subjects as reported in the literature (2.0%), and about one in eight was greater (>2.1%). The effects of these differences on patient comfort, function and performance, as well as the longevity of socket use, remain a topic for future investigation. 10% 0% Acknowledgments 0.0-1.0% 1.1-2.0% 2.1-3.0% 3.1-4.0% 4.1-6.0% Absolute percent volume difference Figure 7. Absolute percent volume difference relative to electronic file shape. The percent age of models and sockets tested that were within different absolute value ranges of the electronic file shape volumes are shown. much error is clinically relevant. How accurate do socket shapes need to be? Part of the difficulty in assessing the clinical impact of socket manufacturing error is that other issues affect fit besides shape. For example, a practitioner can accommodate shape error through use of a thick elastomeric liner, particularly one that flows over time to fill pockets or voids between the limb and socket. Alignment can also be adjusted, or the patient can adapt over time to slight shape imperfections. Thus, to provide clinically useful information into the clinical impact of socket manufacturing error on patient comfort and function, investigations need to be carefully designed, controlled and executed to make sure that only the influence of shape error is assessed. These investigations have yet to be conducted. We can, however, summarize information in the literature related to this topic. Fernie et al.29 measured residual limb volume fluctuations over time on some people with new lower-limb amputation and others with mature lower residual limbs, both trans-tibial and trans-femoral. His clinical observations demonstrated that donning the prosthesis was difficult when there was a volume increase of 3% to 5% within a one-month time interval. Zachariah et al.23 measured absolute residual limb volume changes between sessions 1 to 15 days apart of 1.1% to 2.4% on five subjects and 12.6% on a sixth subject. A 1.1% to 2.4% volume change is consistent with the absolute volume changes of 0.0% to 2.0% measured between morning and afternoon sessions on the same day for eight unilateral trans-tibial amputee subjects.30 In the present study, 53% of models and 60% of sockets had a 1.0% or less difference in absolute volume from the electronic file shape (Figure 7). For the 1.1% to 2.0% range, the results were 37% of the models and 23% of the sockets. The remaining groups, ranging from 2.1% to 6.0% difference from This article is based on the online course titled, ‘‘How Accurate are Carved Positives Made by Central Fabrication Facilities?’’ available through the American Academy of Orthotists and Prosthetists Paul E. Leimkuehler Online Learning Center (OLC) at www.oandp.org/olc. Permission for use of this material for publication in Prosthetic Orthotic International has been granted by the American Academy of Orthotists and Prosthetists. Funding This research was based on work supported by the National Institutes of Health (NIH), National Institute of Bioimaging and Biomedical Engineering grant NIH EB-07329. References 1. O&P Business News. ‘Central Fabrication Resource Guide’, supplement to the 15 May 2009 issue of O&P Business News. 2. Geil MD. Consistency, precision, and accuracy of optical and electromagnetic shape-capturing systems for digital measurement of residual-limb anthropometrics of persons with transtibial amputation. J Rehabil Res Dev 2007; 44(4): 515–524. 3. McGarry T, McHugh B, Buis A and McKay G. Evaluation of the effect of shape on a contemporary CAD system. Prosthet Orthot Int 2008; 32(2): 145–154. 4. Topper AK and Fernie GR. An evaluation of computer aided design of below-knee prosthetic sockets. Prosthet Orthot Int 1990; 14(3): 136–142. 5. Ruder GK. CAD CAM trans-tibial temporary prosthesis: analysis and comparison with an established technique. Prosthet Orthot Int 1992; 16(3): 189–195. 6. Houston VL, Mason CP, Beattie AC, LaBlanc KP, Garbarini M, Lorenze EJ, et al. The VA-Cyberware lower limb prosthetics-orthotics optical laser digitizer. J Rehabil Res Dev 1995; 32(1): 55–73. 7. Steele AL. A survey of clinical CAD/CAM use. J Prosthet Orthot 1994; 6(2): 42–47. 8. Boone DA and Burgess EM. Automated fabrication of mobility aids: clinical demonstration of the UCL computer aided socket design system. J Prosthet Orthot 1989; 1(3): 187–190. 9. Chan RB, Childress DS, Brncick MD, Edwards M and Uellendahl E (1992). Clinical experience with CAD/ CAM below-knee socket fittings. Proceedings 7th World Congress of the International Society for Prosthetics and Orthotics (ISPO), Chicago, IL; 18. Downloaded from poi.sagepub.com by guest on July 17, 2016 89 Sanders et al. 10. Johannesson A, Larsson GU and Oberg T. From major amputation to prosthetic outcome: a prospective study of 190 patients in a defined population. Prosthet Orthot Int 2004; 28(1): 9–21. 11. Smith D and Burgess E. The use of CAD/CAM technology in prosthetics and orthotics: current clinical models and a view to the future. J Rehabil Res Dev 2001; 38(3): 327–334. 12. Prosthetics and Orthotics CAD/CAM Listserver. Responses to a posting dated 25 June 2009. 13. Kohler P, Lindh L and Netz P. Comparison of CADCAM and hand made sockets for PTB prostheses. Prosthet Orthot Int 1989; 13(1): 19–24. 14. Boone DA, Harlan JS and Burgess EM. Automated fabrication of mobility aids: review of the AFMA process and VA/Seattle ShapeMaker software design. J Rehabil Res Dev 1994; 31(1): 42–49. 15. Childress DS. Presentation highlights: computer-aided design and manufacture (CAD-CAM). J Rehabil Res Dev 2002; 39(3 Suppl): 15–16. 16. Engsberg JR, Clynch GS, Lee AG, Allan JS and Harder JA. A CAD/CAM method for custom below-knee sockets. Prosthet Orthot Int 1992; 16(3): 183–188. 17. Brncick M. Computer automated design and computer automated manufacture. Phys Med Rehabil Clin N Am 2000; 11(3): 701–713. 18. Oberg T, Lilja M, Johansson T and Karszina A. Clinical evaluation of trans-tibial prosthesis sockets: a comparison between CAD CAM and conventionally produced sockets. Prosthet Orthot Int 1993; 17(3): 164–171. 19. Sanders JE, Rogers EL, Sorenson EA, Lee GS and Abrahamson DC. CAD/CAM transtibial prosthetic sockets from central fabrication facilities: how accurate are they? J Rehabil Res Dev 2007; 44(3): 395–405. 20. Sanders JE and Severance MR. Measuring foam model shape with a contact digitizer. Prosthet Orthot Int 2011 (in press). 21. Zachariah SG, Sorenson E and Sanders JE. A method for aligning trans-tibial residual limb shapes so as to identify 22. 23. 24. 25. 26. 27. 28. 29. 30. regions of shape change. IEEE Trans Neural Syst Rehabil Eng 2005; 13(4): 551–557. Veltkamp RC and Hagedoorn M. State-of-the-art in shape matching. Utrecht (the Netherlands): Utrecht University, 1999. Zachariah SG, Saxena R, Fergason JR and Sanders JE. Shape and volume change in the trans-tibial residuum over the short-term: preliminary investigation from six subjects. J Rehabil Res Dev 2004; 41: 683–694. Sidles JA, Boone DA, Harlan JS and Burgess EM. Rectification maps: a new method for describing residual limb and socket shapes. J Prosthet Orthot 1989; 1: 149–153. Chahande A, Billakanti S and Walsh N. Lower limb shape characterization using feature extraction techniques: noncontact laser scanning. Conf Proc IEEE Eng Med Biol Soc 1994; 1: 482–483. Borchers RB, Boone DA, Joseph AW, Smith DG and Reiber GB. Numerical comparison of 3-D shapes: potential for application to the insensate foot. J Prosthet Orthot 1995; 7: 29–34. Lemaire ED and Johnson F. A quantitative method for comparing and evaluating manual prosthetic socket modifications. IEEE Trans Neural Syst Rehabil Eng 1996; 4: 303–309. American Board for Certification in Orthotics, Prosthetics, and Pedorthics, Inc, Practice Analysis of Certified Practitioners in the Disciplines of Orthotics and Prosthetics, http://www.abcop.org/certification/ OrthotistsProsthetists/Documents/PracticeAnalysis_SS04. pdf (May 2010). Fernie GR and Holliday PJ. Volume fluctuations in the residual limbs of lower limb amputees. Arch Phys Med Rehabil 1982; 63(4): 162–165. Sanders JE, Zachariah SG, Jacobsen AK, and Fergason JR. Changes in interface pressures and shear stresses over time on trans-tibial amputee subjects ambulating with prosthetic limbs: Comparison of diurnal and six-month differences. J Biomech 2005; 38(8): 1566–1573. Downloaded from poi.sagepub.com by guest on July 17, 2016