COMPARING 3D FOOT SHAPE MODELS BETWEEN TAIWANESE AND JAPANESE FEMALES

Transcription

1 J. Hum. Ergol., 44: 11-20, 2015 COMPARING 3D FOOT SHAPE MODELS BETWEEN TAIWANESE AND JAPANESE FEMALES YU-CHI LEE 1, MAKIKO KOUCHI 2, MASAAKI MOCHIMARU 1 AND MAO-JIUN WANG 1 * 1 Department of Industrial Engineering and Engineering Management, National Tsing Hua University, No. 101, Section 2, Kuang-Fu Road, Hsinchu 30013, Taiwan * mjwang@ie.nthu.edu.tw 2 Digital Human Research Center, National Institute of Advanced Industrial Science and Technology, Aomi, Koto-ku, Tokyo , Japan. ABSTRACT This study compares foot shape and foot dimensions between Taiwanese and Japanese females. One hundred Taiwanese and 100 Japanese female 3D foot scanning data were used for comparison. To avoid the allometry effect, data from 23 Taiwanese and 19 Japanese with foot length between 233 to 237 mm were used for shape comparison. Homologous models created for the right feet of the 42 subjects were analyzed by Multidimensional Scaling. The results showed that there were significant differences in the forefoot shape between the two groups, and Taiwanese females had slightly wider feet with straighter big toe than Japanese females. The results of body and foot dimension comparison indicated that Taiwanese females were taller, heavier and had larger feet than Japanese females, while Japanese females had significantly larger toe 1 angle. Since some Taiwanese shoemakers adopt the Japanese shoe sizing system for making shoes, appropriateness of the shoe sizing system was also discussed. The present results provide very useful information for improving shoe last design and footwear fit for Taiwanese females. Key words: Homologous modelling; foot anthropometry; ethnic differences; shoe sizing INTRODUCTION Human feet have complicated anatomical structure and geometric features. Different ethnic groups tend to have different foot shape characteristics. Previous studies applied some foot feature variables to classify foot shape types (Kouchi, 1995; Mochimaru et al., 2000; Nilsson et al., 2012; Rodrigo et al., 2012). Krauss et al. (2008) classified European feet into three types including voluminous, flat and slender. Luo et al. (2009) reported that American females had narrow and high arches and American males had broad and low arch foot shapes. Hong et al. (2011) found that among the Chinese people, short feet tended to be associated with voluminous feet, and narrow and flat feet tended to be associated with long feet. These studies reported special findings in foot characteristics based on their own foot anthropometric data and provided useful information for footwear design (Sun et al., 2009). For global footwear manufacturing and marketing, it is important to understand the ethnic differences in foot shape and take these differences into consideration in footwear design and production. Hawes et al. (1994) compared the two-dimensional forefoot shape differences between different ethnic groups using a database of 708 American and 513 Asian (Japanese and Korean) male subjects. They found that the Asian male s feet had a less acute angle in the long axis than that of the North American and the difference in forefoot shape was due to foot breadth difference. Kusumoto et al. (1996) compared the ethnic differences in foot morphology between Filipino and Japanese females. They found a significant difference in the toe 1 angle between the two groups. Gonda and Katayama (2006) reported that the Tongans had significantly longer and wider feet than the Japanese, French, Australian Aborigines and West Africans. Received 20 February 2015; accepted 12 June 2015

2 12 Y.- C. LEE et al. More recently, Lee and Wang (2014) classified the Taiwanese female s feet into three foot types including NL (narrow foot with low arch height), SH (short instep length, with high arch height) and LW (long instep length with wider foot and average arch height) types and indicated that instep length, foot breadth and arch height were the three principal foot dimensions for classifying Taiwanese female foot shapes. In real world practice, some Taiwanese shoemakers used the Japanese shoe sizing system (JIS S- 5037, 1998) as a basic reference, and made some modifications based on their own experiences. It seems that there is a lack of information about the foot shape differences between the two groups. Using the shoe last design with inadequate shape and an inadequate sizing system may lead to a poor fitting problem. Hence, this study aims to compare foot shape between Taiwanese and Japanese females by analyzing 3D foot homologous models and provide some specific information to the practitioners. METHODS Subjects Three dimensional foot scanning data of 100 Taiwanese females and 100 Japanese females were randomly sampled from the foot database of 1000 Taiwanese and 410 Japanese. The Taiwanese data were collected in the Digital Ergonomics Laboratory, National Tsing Hua University, Taiwan. The Japanese data were collected in the Digital Human Research Center, National Institute of Advanced Industrial Science and Technology (AIST), Japan. None of the subjects had any history of visible foot abnormalities or foot illnesses. The 3D scanning data were collected from February 2011 to June The protocol for Taiwanese data collection was approved by the Institutional Ethics Committee of National Tsing Hua University. The Japanese experimental protocol was reviewed and approved by an Institutional Ethics Committee of AIST. The basic demographic data and foot dimensions of the Taiwanese and Japanese female subjects are presented in Table 1. The average age was 22.2 years for the Taiwanese females and 23.7 years for the Japanese females. The mean height and weight was mm and 52.0 kg for the Taiwanese females, and was mm and 49.2 kg for the Japanese females, respectively. Table 1. The means and standard deviations of the body dimensions of the Japanese and Taiwanese females and the t-test results.

3 COMPARING TAIWANESE AND JAPANESE FEMALE FOOT SHAPES 13 Measurement system The same type of 3D foot scanner (INFOOT USB scanning system, IFU-S01, I-ware Laboratory Co., Ltd, Japan), was used to obtain the 3D foot images for both the Taiwanese and Japanese females. The scanning volume was L400 x B200 x H150 (mm), with a 10 second scanning duration. The foot scanner accuracy was within 1.0 mm. The same training program was provided by the same scanner vendor to train the experimenters of the two institutes. Before scanning, special markers were pasted onto the metatarsale tibiale (the most medial point of the head of the first metatarsale bone), the metatarsale fibulare (the most lateral point of the head of the fifth metatarsale bone), the sphyrion (the lowest point of the medial malleolus of the tibia), the sphyrion fibulare (the lowest point of the lateral malleolus of the fibula), and the tentative junction point (crossing point of the tendon to the fifth toe and the crease between the leg and the foot). The foot scanning system could automatically detect the landmark positions and calculate the 14 foot dimensions. Positions of the pternion (the rearmost point of the heel) and the tips of the first and second toes were automatically identified. The scan-derived foot dimensions are comparable to measurements taken by an expert (Kouchi and Mochimaru, 2001). The definitions of the 14 foot dimensions are illustrated in Fig JP 1 MT 4 5 Foot axis 2 MF 11 9 Sph. f Pte LP 50% FL 8 7 MF 12 3 Sph %FL Pte MT 8 50% FL Pte LP Fig. 1. The 14 scan-derived foot dimensions. The numbers are the same with those in Table 1. FL: foot length, JP: junction point, LP: landing point, MF: metatarsale fibulare, MT: metatarsale tibiale, Pte: pternion, Sph: sphyrion, Sph. f.: sphyrion fibulare. See text for the definitions of landmarks and measurements. Experiment procedures Before scanning, each subject was requested to clean her feet using wet tissues. A well-trained research assistant placed special markers onto the five landmark positions. Both feet were scanned after positioning the landmarks. The subject was instructed to stand naturally, looking straight ahead, with their weight distributed equally on both feet. The noise was removed manually from each image after the scan. The procedure followed the previous studies of Mochimaru et al. (2000) and Kouchi and Mochimaru (2001). Before the analysis, one research assistant examined the detected landmark positions of all 200 scanning images (Taiwanese and Japanese) and the obvious errors were corrected. This procedure could minimize inter-observer errors in data collection. Homologous modelling Subjects with the same foot length range were selected for comparison to avoid the effects of allometry (Anil et al., 1997; Wunderlich and Cavanagh, 2001; Voracek et al., 2007; Krauss et al., 2008; Hong et al., 2011). Figure 2 shows the foot length (FL) size distribution for the 100 Taiwanese females and 100 Japanese females. The largest group belongs to the FL size of 235 mm (233 to 237 mm) for the Taiwanese females, and 230 mm (228 to 232 mm) for the Japanese females. Subjects with foot length ranging from 233 to 237 mm were selected because the number of subjects for the two groups was close. A total of 42 subjects (19 Japanese and 23 Taiwanese) were selected, and their data were used for foot shape comparison. A homologous model was created for the right foot of each of the 42 subjects to compare the three-

4 14 Y.- C. LEE et al Number Taiwanese females Japanese females Foot length (mm) Fig. 2. Foot length distribution of the Japanese and Taiwanese females. Fig. 3. The process of homologous modelling. A: original scan and measured landmarks, B: homologous model created using Di+ and data points used as landmarks, C: final homologous model created by template fitting using mhbm, D: template and landmarks. dimensional foot shape quantitatively (Mochimaru et al., 2000). The homologous foot models were generated using a template fitting method (Yamazaki et al., 2013). Figure 3 shows the homologous modeling process. Figure 3A shows the original scan and the measured landmarks. Four of the measured landmarks and 184 data points created by Di+ (I-ware Laboratory Co., Ltd.) were used as landmarks in template fitting (Fig. 3B). A polygonal mesh model with 189 landmarks, as shown in Fig. 3D was used as the template. Subsequently, the template was fitted to each scanned data based on the 189 landmarks using mhbm software (Medic Engineering, Inc.). The foot coordinate system was defined as follows: XY plane is the standing surface, and the pternion projected onto the XY plane was used as the origin. The X axis is identical with the foot axis, the line connecting the pternion and the centre of the vertical cross section passing the metatarsale tibiale and the metatarsale fibulare. The rightward direction is Y+ and the upward direction is Z+. Statistical analysis The right foot data were used for statistical analyses. The distance between two homologous models was calculated as the sum of the distances between all corresponding data points of the two homologous models. The obtained distance matrix was analyzed using the multi-dimensional scaling method (MDS) (Cox and Cox, 1994) by HBS software (Digital Human Technology Inc.) (Mochimaru and Kouchi, 2000). Using MDS, the distance relationships between the subjects were summarized in a low-dimensional space. Each dimension represented a variation in shape and was independent of other dimensions. The fit of the model was evaluated using the squared correlation coefficient (RSQ) between the elements of the original

5 COMPARING TAIWANESE AND JAPANESE FEMALE FOOT SHAPES 15 distance matrix and the elements of the distance matrix calculated from the obtained low-dimensional space. In addition, the significance of the between-group differences in the mean scores was tested using a t-test for each dimension. Basic statistics were obtained for the 14 scan-derived foot dimensions. The difference between Taiwanese and Japanese female foot dimensions was tested using the t-test. The significance of the differences between 100 Taiwanese and 100 Japanese females as well as between 23 Taiwanese and 19 Japanese females with the same FL size was tested. A liner regression equation between the ball girth and foot length was calculated for each group. These two variables were chosen to facilitate comparisons with the current shoe sizing system constructed by using these variables. All statistical analyses were conducted using SPSS 16.0 software package. The significance level was set at α= RESULTS AND DISCUSSION Foot shape A three-dimensional solution of the multi-dimensional scaling (MDS) method was obtained with Stress = 0.14, and R 2 = In order to interpret the shape variation represented by the three dimensions, virtual foot shapes with score of ± 3 standard deviation for each dimension were created. Table 2 shows the virtual foot shapes and the t-test results. The first dimension represents the variation in size. The second dimension characterizes the arch of the foot. The third dimension is related to the forefoot shape. A significant between-group difference was observed only in the third dimension. Figure 4 shows the differences between virtual forefoot shapes at ±3 standard deviation in the third dimension of MDS results. Since the Taiwanese females had significantly smaller scores than Japanese females, results indicated that Japanese females tended to have greater toe 1 angle, as well as greater length between the pternion and the tip of the big toe due to a greater instep length. Taiwanese females tended to have wider foot breadth than the Japanese females. The results revealed a significant difference in forefoot shape between the two groups. Table 3 shows basic statistics of the two groups of the same foot length size, 235 mm, and the results of t-test. Significant difference was found only in the toe 1 angle. Figure 5 illustrates the outlines of the average of the 23 Taiwanese and 19 Japanese females feet. The foot outline of the Taiwanese females was slightly wider at the toe 1 area than that of the Japanese females, and was slightly shorter than that of the Table 2. ANOVA results with virtual foot shape at 3 standard deviation (thick black line) and +3 standard deviation (thin grey line) on each MDS dimension.

6 16 Y.- C. LEE et al. The 3 rd dimension Forfoot shape differences Fig. 4. The differences in the forefoot area represented by the 3rd dimension of the MDS results. Virtual foot shape at 3 standard deviation (thin line) and +3 standard deviation (thick line). The feet of Japanese females are closer to the foot in thick line and Taiwanese females are closer to the foot in thin line. Table 3. The means, standard deviations and t-test results in comparing Japanese and Taiwanese females foot dimensions in the same foot length. Upper confidence Lower confidence ID Measurement item Japanese (n=19) Taiwanese (n=23) Difference t-test interval interval Age (year) 23.2± ± n.s. Stature (mm) ± ± n.s. Weight (kg) 50.2± ± n.s. BMI (kg/m 2 ) 19.3± ± n.s. 1 Foot length (mm) 235.2± ± n.s. 2 Instep length (mm) 173.3± ± n.s. 3 Fibular instep length (mm) 153.3± ± n.s. 4 Foot breadth diagonal (mm) 93.8± ± n.s. 5 Foot breadth horizontal (mm) 89.9± ± n.s. 6 Heel breadth (mm) 59.2± ± n.s. 7 Ball girth (mm) 223.5± ± n.s. 8 Instep girth (mm) 221.3± ± n.s. 9 Heel girth (mm) 294.7± ± n.s. 10 Sphyrion height (mm) 60.4± ± n.s. 11 Fibular sphyrion height (mm) 47.2± ± n.s. 12 Instep height (mm) 57.2± ± n.s. 13 Toe 1 angle ( ) 13.0± ± * 14 Toe 5 angle ( ) 10.0± ± n.s. : Japanese Taiwanese; *: p < 0.05; n.s.: not significant Japanese females. The shape difference was also observed in the toe 1 angle in two groups. The Taiwanese females had a straighter big toe than Japanese females. These features suggest that the Taiwanese females forefoot shape tends to be square. The Japanese females had a more abducted and longer big toe than Taiwanese females. Thus, the Japanese females forefoot shape tended to be slender. On the other hand, Table 3 indicates that there were no significant differences in length and breadth measurements between Japanese and Taiwanese females in the same foot length range. The forefoot shape difference was characterized by toe 1 angle. The finding was consistent with the results of shape analysis using MDS. The means and standard deviations of the 14 scan-derived foot dimensions of the 100 Taiwanese and

7 COMPARING TAIWANESE AND JAPANESE FEMALE FOOT SHAPES 17 Taiwanese Japanese Fig. 5. The outlines of the average foot models of the 23 Taiwanese and 19 Japanese females with foot length size of 235 mm. 100 Japanese females are presented in Table 1. The Taiwanese females were taller and heavier and had larger feet than Japanese females. Taiwanese females had significantly longer foot length (p<0.001), wider heel (p<0.05), and greater instep girth, heel girth, and instep height (p<0.05) than those of the Japanese females. However, no difference was observed in foot breadth, ball girth, sphyrion height and fibular sphyrion height. The toe 1 angle of the Japanese females was greater than that of the Taiwanese females by 1.6 (Table 1). This is perhaps due to the cultural differences in shoe wearing. Wearing different footwear types would influence foot dimensions and shapes, especially in forefoot area (Ashizawa et al., 1997). Japanese and Taiwanese children usually have to wear uniform shoes from elementary school to high school. The uniform shoes design has varieties chosen by the schools concerned and not nationally standardized. Students choose the best shoes by trying on the shoe for individuals. Since the uniform shoes do not have ball girth size variations, children must wear shoes that do not fit their feet well both in foot length and ball girth in Japan. This may cause the difference in the toe 1 angle observed between Taiwanese and Japanese females. However, It is difficult to clarify the difference of two groups in genetic and environmental factors. Among the differences in foot shape, especially the difference of the toe 1 angle and the toe 5 angle, it is necessary to investigate in further studies. Shoe sizing system Figure 6 illustrates the distribution of the right feet of subjects based on the foot length (FL) (x-axis) Shose sizing system width E 240 Ball girth (mm) R = R = Taiwanese females Japanese females Regression: Taiwanese female Regression: Japanese female Foot length (mm) Fig. 6. The distributions and regression lines of the Japanese and Taiwanese females in comparison with the regression lines of the current shoe sizing system.

8 18 Y.- C. LEE et al. Table 4. Results of liner regression equation of Taiwanese and Japanese females. and ball girth (BG) (y-axis), as well as the liner regression lines of the two groups. The R square value was and for the Taiwanese and Japanese females, respectively. A significant difference was found between the two regression lines (p<0.05). Figure 6 also shows the line representing the relationship between the FL and BG in the BG size E, the most commonly observed width, in the Taiwanese as well as Japanese shoe sizing system. According to the grading rule in shoe sizing system, an interval increase of 3 mm in BG when FL has an increase of 5 mm, the slope of the line is 0.6, thus the regression line is BG = FL. With the increase or decrease of one BG size, the y intercept changes by 6 mm. The liner regression models of the two groups are shown in Table 4. The regression coefficient (slope) was 0.63 for Taiwanese females, and 0.49 for Japanese females. The 95% confidence interval of the regression coefficient was 0.47 to 0.81 for Taiwanese females and 0.30 to 0.68 for Japanese females. The slope of the sizing system, 0.6, was within the 95% confidence interval of the regression coefficient for the two regression lines. Thus, the regression results of Taiwanese and Japanese females were not significantly different in the grading slope of the two shoe sizing systems. The current shoe sizing system in Taiwan, i.e. CNS 4800-S1093 (2000) specifies that the interval between two sizes is 5 mm for FL size and 6 mm for BG size. For ladies shoes, FL size ranges from 195 mm to 270 mm, and BG size ranges from A to F (183 mm to 276 mm). The rule and the size range of FL size are the same in the Japanese shoe sizing system (JIS S-5037) for adult females, but the BG size range is from A to G (183 mm to 282 mm) for the Japanese (JIS S-5037, 1998). Figure 7 illustrates the shoe sizes of CNS 4800-S1093 and the regression lines of the Taiwanese and Japanese females. There are nine BG sizes for females, i.e. A, B, C, D, E, EE, EEE, EEEE and F. For the sizing system, the medium BG size was E, and the difference between two BG sizes was 6 mm. The regression line (regression coefficient = 0.49) for the Japanese females in Fig. 7 shows that the line passes BG size E when FL is 240 and smaller, but it passes BG size D when FL is 250 and larger. The regression line for the Taiwanese females with regression coefficient of 0.63 was very close to 0.6. But, the line passes BG size D. This means that BG size D is closer to the average Taiwanese feet. Generally, not all shoe sizes shown in Fig. 7 are manufactured. Only the sizes having more customers are manufactured. In Japan, it is very difficult to find shoes with FL size smaller than 220 mm or greater than 245 mm, or for BG size D or narrower, or for size EEE or wider. In Taiwan, FL size in commercially available shoes ranged from 235 to 250 mm, and BG size range was the same as in Japan. Inconsistency between the Japanese shoe sizing system and the present Japanese subjects may be explained by the intergeneration differences in foot sizes due to the secular changes (Kouchi, 1998). On the other hand, the possible reasons that the Taiwanese females tend to have difficulty in finding properly fitting shoes may be due to the discrepancy of the size distribution range between commercially available shoes and Taiwanese female foot sizes. Since the mean foot length of the Taiwanese females was about 235 mm and BG size D, the commercially available shoes may accommodate only half of the young adult Taiwanese females. The manufacturing of wider FL size range and BG size range shoes can help to solve the problems.

9 COMPARING TAIWANESE AND JAPANESE FEMALE FOOT SHAPES 19 Ball girth (mm) BG = FL BG = FL Foot length (mm) F EEEE EEE EE E D C B A Regression: Taiwanese females Regression: Japanese females Fig. 7. The CNS 4800-S1093 shoe sizing system with regression lines for Taiwanese (broken line) and Japanese (solid line) young adult females. CONCLUSIONS This study compared the foot shape between Taiwanese and Japanese females of the same FL size and found a significant difference in the forefoot shape. Since some of the Taiwanese shoemakers refer to the Japanese shoe last and make slight modifications, the shoe fit problem may be encountered if the differences in foot shape between the two groups are not recognized. Thus, it is necessary to modify the forefoot shape particularly in the toe 1 angle of the last to improve the shoe fit for the Taiwanese females. Further, when comparing the foot size distribution with the current shoe sizing system, the young adult Taiwanese females feet are narrower than the average BG size in the shoe sizing system, and the range of FL sizes of commercially available shoes are too small. Thus, the change in shoe manufacturing is necessary to increase both FL sizes and BG sizes so that more customers can be accommodated. Overall, the 3D foot shape comparisons obtained in this study can provide very useful information for designing the shoe last or sizing system for Taiwanese females to facilitate fit wearing. ACKNOWLEDGEMENT The authors would like to express our great appreciation to the financial support by the National Science Council under the Grant No. NSC I REFERENCES Anil, A, Peker, T, Turgut, HB and Ulukent, SC (1997) An examination of the relationship between foot length, foot breath, ball girth, height and weight of Turkish university students aged between 17 and 25. Anthropol. Anz., 55: Ashizawa, K. Kumakura, C, Kusumoto, A and Narasaki, S (1997) Relative foot size and shape to general body size in Javanese, Filipinas and Japanese with special reference to habitual footwear types. Ann. Hum. Biol., 24:

10 20 Y.- C. LEE et al. CNS 4800-S1093 (2000) Sizing system for shoes, Taiwan. Cox, TF and Cox, MAA (1994) Multidimensional Scaling, London: Chapman & Hall. Gonda, E and Katayama, K (2006) Big feet in Polynesia: a somatometric study of the Tongans. Anthropol. Sci., 114: Hawes, MR, Sovak, D, Miyashita, M, Kang, SJ, Yoshihuku, Y and Tanaka, S (1994) Ethnic differences in forefoot shape and the determination of shoe comfort. Ergonomics, 37: Hong, Y, Wang, L, Xu, DQ and Li, JX (2011) Gender differences in foot shape: a study of Chinese young adults. Sport Biomech., 10: JIS S-5037 (1998) Sizing system for shoes, Japanese industrial standards, Japan. Kouchi, M (1995) Analysis of foot shape variation based on the medial axis of foot outline. Ergonomics, 38: Kouchi, M (1998) Foot dimensions and foot shape: differences due to growth, generation and ethnic origin. Anthropol. Sci., 106: Kouchi, M and Mochimaru, M (2001) Development of a low cost foot-scanner for a custom shoe making system. Proceedings of the 5th Symposium on Footwear Biomechanics, Zurich, Switzerland. Krauss, I, Grau, S, Mauch, M, Maiwald, C and Horstmann, T (2008) Sex-related differences in foot shape. Ergonomics, 51: Kusumoto, A, Suzuki, T, Kumakura, C and Ashizawa, K (1996) A comparative study of foot morphology between Filipino and Japanese women, with reference to the significance of a deformity like hallux valgus as a normal variation. Ann. Hum. Biol., 23: Lee, YC and Wang, MJ (2015) Taiwanese adult foot shape classification using 3D scanning data. Ergonomics, DOI: 58: Luo, G, Houston, VL, Mussman, M, Garbarini, M, Beattie, AC and Thongpop, C (2009) Comparison of male and females foot shape. J. Am. Podiatr. Med. Assoc., 99: Mochimaru, M and Kouchi, M (2000) Statistics for 3D Human Body Forms. SAE Technical Report, 2000: Mochimaru, M, Kouchi, M and Dohi, M (2000) Analysis of 3-D human foot forms using the free form deformation method and its application in grading shoe lasts. Ergonomics, 43: Nilsson, MK, Friis, R, Michaelsen, MS, Jakobsen, PA and Nielsen, RO (2012) Classification of the height and flexibility of the medial longitudinal arch of the foot. J. Foot Ankle Res., 5:3 DOI: / Rodrigo, AS, Goonetilleke, RS and Witana, CP (2012) Model based foot shape classification using 2D foot outlines. Comput. Aided Des., 44: Sun, SP, Chou, YJ and Sue, CC (2009) Classification and mass production technique for three-quarter shoe insoles using non-weight-bearing plantar shapes. Appl. Ergon., 40: Voracek, M, Fisher, ML, Rupp, B, Lucas, D and Fessler, DM (2007) Sex differences in relative foot length and perceived attractiveness of female feet: relationships among anthropometry, physique, and preference ratings. Percept. Motor Skills, 104: Wunderlich, RE and Cavanagh, PR (2001) Gender differences in adult foot shape: implication for shoe design. Med. Sci. Sports Exerc., 33: Yamazaki, S, Mochimaru, M and Kouchi, M (2013) Markerless landmark localization on body shape scans by non-rigid model fitting. Proceedings of the 2nd International Digital Human Modeling Symposium, Michigan, USA.