Personalized Body Segment Parameters From ... - Semantic Scholar

1756

IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 52, NO. 10, OCTOBER 2005

Personalized Body Segment Parameters From Biplanar Low-Dose Radiography Raphaël Dumas, Rachid Aissaoui*, Member, IEEE, David Mitton, Wafa Skalli, and Jacques A. de Guise

Abstract—Body segment parameters are essential data in biomechanics. They are usually computed with population-specific predictive equations from literature. Recently, medical imaging and video-based methods were also reported for personalized computation. However, these methods present limitations: some of them provide only two-dimensional measurements or external measurements, others require a lot of tomographic images for a three-dimensional (3-D) reconstruction. Therefore, an original method is proposed to compute personalized body segment parameters from biplanar radiography. Simultaneous low-dose frontal and sagittal radiographs were obtained with EOS™ system. The upper leg segments of eight young males and eight young females were studied. The personalized parameters computed from the biplanar radiographic 3-D reconstructions were compared to literature. The biplanar radiographic method was consistent with predictive equations based on -ray scan and dual energy X-ray absorptiometry. Index Terms—Biomechanics, biplanar radiography, body segment parameters, personalization, prediction methods, 3-D reconstruction.

I. INTRODUCTION

T

HE BODY segment parameters (BSP) are essential data for the study of human movement and posture in sports, ergonomics, rehabilitation, and orthopedics. The segment mass, the position of the center of mass, and the inertia tensor are for example necessary in biomechanics for the computation of inverse dynamics. The BSP are usually obtained from several predictive equations reported in the literature [1]–[13] with scaling or multiple regression. Some predictive equations are based on cadaveric measurements [2]–[4] and relate to small samples adult Caucasian able-bodied males. Other predictive equations are based Manuscript received July, 28 2004; revised November 28, 2004. This work was supported in part the National Science and Engineering Research Council of Canada, in part by the Fonds Québécois pour la Recherche en Nature et Technologie, in part by the Ministère Français des Affaires Etrangères (Lavoisier programme), in part by the Canadian Institutes of Health Research (Mentor programme), in part by the Ministère Français de la Recherche (Fond National pour la Science), and in part by the Région Ile de France (Sesame programme). Asterisk indicates corresponding author. R. Dumas was with the Laboratoire de recherche en imagerie et orthopédie, Ecole de Technologie Supérieure, Montreal, Montreal, QC H3C 1K3, Canada. He is now with the Laboratoire de Biomécanique et Modélisation Humaine, INRETS—Université Claude Bernard, 69622 Lyon, France. *R. Aissaoui is with the Laboratoire de recherche en imagerie et orthopédie, Ecole de Technologie Supérieure Montreal, Montreal, QC H3C 1K3, Canada (e-mail: [email protected]). D. Mitton and W. Skalli are with the Laboratoire de Biomécanique, CNRS UMR 8005—Ecole Nationale Supérieure d’Arts et Métiers, 75013 Paris, France. J. A. de Guise is with the Laboratoire de recherche en imagerie et orthopédie, Ecole de Technologie Supérieure Montreal, Montreal, QC H3C 1K3, Canada. Digital Object Identifier 10.1109/TBME.2005.855711

on living subjects using stereo-photogrammetry [1], [5]–[7], -ray scan [8] and, more recently, dual energy X-ray absorptiometry (DEXA) [9], [10]. The studies on living subjects deal in general with a larger sample and different populations such as able-bodied children, adolescents and women. Moreover, there are significant differences between predictive equations [9], [14]. Thus, choosing the appropriate predictive equation remain very difficult [15], [16] especially for pathologic populations. That is why medical imaging methods such as DEXA [9], [10], magnetic resonance imaging (MRI) [17], [18], computed tomography (CT) [19], and video scan [20], [21] were recently proposed. These methods were shown to be effective in measuring personalized BSP. However these methods remain limited: some of them provide only two-dimensional (2-D) measurements (DEXA) or external measurements (video scan), others require a lot of tomographic images for a three-dimensional (3-D) reconstruction (MRI and CT). Especially, 3-D reconstructions from MRI and CT yield time, cost and radiation considerations. As a alternative, the biplanar radiographic method is a recent 3-D imaging method already used in orthopedics for the personalized 3-D reconstructions of the skeleton [22]–[26]. The mean accuracy of the 3-D reconstructions was reported to be about 1 mm for vertebrae, pelvis, and femur [26]–[32]. The objective of this study is to determine personalized BSP using a biplanar radiographic method. This paper presents an application of the method for the upper leg segment and a comparison of the personalized BSP to the literature. II. MATERIALS AND METHODS A. Subjects and eight ableEight able-bodied males participate in this study. The bodied females cm for males and cm for females. height was The weight was for males and for females. Informed consent was obtained from each subject. B. Biplanar Low Dose Radiography Frontal and sagittal numerical radiographs were taken at the Laboratoire de Biomécanique with EOS™ system (Biospace, Paris, France). EOS™ system [Fig. 1(a)] includes two X-ray sources and two numerical detectors. The detection technology is derived from G. Charpak’s research work (Physics Nobel Prize 1992). In the perspective of BSP computation for the upper leg segment, two simultaneous lower limb low-dose radiographs were obtained [Fig. 1(b)]. The radiographs were centered in order to visualize the full right leg. The characteristic of these biplanar

0018-9294/$20.00 © 2005 IEEE

DUMAS et al.: PERSONALIZED BSPs FROM BIPLANAR LOW-DOSE RADIOGRAPHY

1757

Fig. 2. Digitization of the 2-D anatomical points and contours using a specific software (collaboration between the Laboratoire de recherche en imagerie et orthopédie and the Laboratoire de Biomécanique).

Fig. 1. (a) EOS™ system; (b) Biplanar low-dose radiographs of the lower limb.

radiographs is that they are taken in a calibrated radiological environment. This allows a 3-D reconstruction of internal and external surfaces as presented below. For this purpose, a global coordinate system is defined according to the axes of the EOS™ system. C. Three Dimensional Reconstruction 1) Three-Dimensional Reconstruction Technique: The 3-D personalized surface models of femur, patella, and thigh were obtained by biplanar radiographic method using nonstereo corresponding contour (NSCC) [28]. The NSCC method has been completely presented and discussed elsewhere [28]. To summarize, the method is based on the registration and deformation of anatomical atlas surface models on 2-D anatomical points and contours that are digitized in both radiographs. Therefore, to be applied to the upper leg segment, this method requires anatomical atlas surface models of femur, patella, and thigh. At this point, the choice of segment boundaries was introduced. Following an anatomical segmentation [33], the entire femur and patella were included in the upper leg segment. Thus, the anatomical atlas surface models of right femur and patella

Fig. 3. Personalized 3-D reconstruction of the upper leg segment, Three-dimensional anatomical points and segment coordinate system.

were obtained from CT of entire dried bone (SliceOmatic, TomoVision, Canada). Besides, the 3-D surface model of the complete right leg of a male living subject was obtained from video scan (3D capturor II, InSpeck, Canada). The thigh anatomical atlas surface model was extracted following the anatomical segmentation [33] at hip and knee joints. At the end, all models were polyhedral closed surfaces with outward normals. The 2-D digitization and the 3-D reconstruction were performed with a specific software developed in collaboration between the Laboratoire de recherche en imagerie et orthopédie and the Laboratoire de Biomécanique. The 2-D digitization of anatomical points and contours was directly achieved onto the numerical radiographs (Fig. 2). The contours were extracted semi-automatically by means of active contour algorithms. The digitization time was about 15 min. As an output of the software, the 3-D reconstruction provided the personalized 3-D surface ), patella (noted ), and femur models of the thigh (noted ), expressed in the global coordinate system (Fig. 3). (noted

1758


2) Three-Dimensional Anatomical Points, Segment Length and Segment Coordinate System: In the perspective of BSP computation, several 3-D anatomical points were also extracted and . These 3-D from the personalized 3-D surfaces anatomical points are widely used in biomechanics [4], [9], [33]–[36] and allow to define the segment length and the segment coordinate system. Some of these 3-D anatomical points were internal landmarks: center of femoral head (FH) and center of posterior part of medial and lateral femoral condyles (MC, LC). These points were computed as center of best fitted . Other 3-D anatomical sphere with a subset of points from points were external landmarks: medial and lateral points of hip boundary (MH, LH), medial and lateral femoral epicondyles (ME, LE), and medial and lateral points of knee boundary (MK, LK). MH corresponds to the perineum, MK and LK correspond to the ridges of the tibial plateau. These points were identified . on Thus, the upper leg segment length could be defined by the distance between FH and the middle of ME and LE. As well, a segment coordinate system could be constructed from these 3-D anatomical points. According to the recommendations of the International Society of Biomechanics [36] for the upper leg seg-axis is the vector from the middle of ME and LE to ment, FH. -axis is the vector normal to the plane containing the -axis three previous internal landmarks pointing anteriorly. and axes. The origin is fixed is the cross product of at FH. In order to compute the personalized BSP of upper leg seg, , and were transformed from the global coorment, dinate system into the segment coordinate system.

The inertial parameters of the upper leg segment were com, puted from the volume characteristics of each parts ( , ). To do so, the average densities of bone tissue and soft cm , tissue were assumed to be: cm for males and cm for females [38]. Thus, mass of the upper leg segment (noted ) was computed by

D. Personalized BSP

1) Mean BSP: The mean BSP (respectively, for the eight males and eight females) obtained with the present method were compared to a study of Chandler et al. on six male cadavers [2]. This cadaver study was chosen because it gives access to the 3-D position of center of mass and the 9 9 inertia tensor in a coordinate system which is consistent with the segment co, , ) defined above (see ordinate system (LH, Section II-C-II). 2) Predictive Equations: The present method was compared also with five predictive equations based on cadaveric measurements [2], [4] and medical imaging methods [9], [11], [17]. These five predictive equations provide the segment mass rel, the transverse position of center atively to the body mass and, the of mass relatively to the segment length moments of inertia expressed as radii of gyration relatively to , , and ). These five the segment length ( predictive equations were chosen because the present study is

, , and were As personalized 3-D surface models obtained, the methodology was to consider the volumes enclosed by these surfaces, to deduce the volumes of soft and bone tissues and to apply the densities of soft and bone tissues. enclosed in was noted : the internal The part of femur surface. By integration on the polyhedral surface models [37], the volume , the position of the center , and the volume inertia tensor at the origin of volume were computed. . Besides, the volume of One my notice that: soft tissue and bone tissue were given by (1) and (2)

(3) The position of the center of mass of the upper leg segment ) was computed by (noted

(4) The mass inertia tensor of upper leg segment at the origin ) was computed by (noted

(5) The mass inertia tensor of upper leg segment at the center of ) was computed by the parallel axis theorem, mass (noted shown in (6) at the bottom of the page, with , , and coordinate indexes. E. Comparison of Personalized BSP to Literature

(6)


TABLE I MEAN (SD) VALUES OF THE POSITION CENTER OF MASS AND INERTIA TENSOR IN THE SEGMENT COORDINATE SYSTEM

TABLE II PREDICTIVE EQUATIONS FOR MALES AND FEMALES BASED ON THE PRESENT METHOD AND COMPARED TO THE LITERATURE

anticipated to disagree with the cadaveric measurements and agree with the other medical imaging methods. While comparing the BSP from the present study and from these five predictive equations, the frontal, sagittal, and transverse axes of the segment were assumed to be equivalent from , , one study to another (respectively, consistent with ). First, the BSP obtained with the present method were and computed relatively to the subject mass and segment length in order to be compared with the five predictive equations. Second, the five predictive equations were used to estimate the BSP ( , , , , and ) for every subject of the present study. To do so, the differences of segment endpoints definition were taken into account. The segment endpoints of de Leva [11], [39] and Chandler et al. [2] are defined by FH and the middle of ME and FE. The segment endpoints of Cheng et al. [17] and Dempster [4] are defined by FH and the middle of

1759

MC and LC. The segment endpoints of Durkin et al. [9] are defined by the middle of MH and LH and the middle of MK and , , and were estimated LK. Therefore, from the radii of gyration using the respective endpoint-to-endpoint segment length. Also, the center of mass was estimated in the segment coordinate system aligned with the 3-D anatomical points, respectively, defined as endpoints. The -coordinate of the center of mass was considered to be . Moreover, different predictive equations for males and females were provided by de Leva [11] and Durkin et al. [9], but not by Chandler et al. [2], Dempster [4], and Cheng et al. [17]. Though, these tree last predictive equations were also used for females. In order to indirectly validate the results of present study compared to the five predictive equations, significant differences were assessed (Statistica, StatSoft, Tulsa, OK). Analysis of variance (ANOVA) followed by Scheffé post hoc tests were performed separately on the five dependent variables ( , , , , and ) considering two factors: method and gender. III. RESULTS Table I presents the mean and standard deviation of the BSP obtained by the present method and that of Chandler et al. [2]. The values were different but shows the same tendency: the center of mass was found slightly medial and posterior and the inertia tensor was found not principal in the segment coordinate system. Table II presents the predictive equations derived from the present method and compared to the five predictive equations from literature. Although differences of method and population, all the predictive equations were fairly consistent except for the segment mass of females that seems somewhat larger with the present method. , , The comparison of the BSP ( , , and ) obtained by the present method and estimated from the predictive equations are presented Fig. 4(a)–(e). Concerning the segment mass , ANOVA tests and showed a significant effect for both method gender . There was no interaction between method and gender effects. The present method showed significant differences (Scheffé post hoc) with predictive equations from cadaveric measurements [2], [4] and DEXA [9]. Concerning the , ANOVA tests transverse position of center of mass showed significant effects for method and gender and a significant interaction between method and gender ef. The present method showed significant fects differences (Scheffé post hoc) with predictive equations from cadaveric measurements [2], -ray scan [11] and MRI [17]. Concerning the frontal and sagittal moments of inertia and , ANOVA tests showed significant effects and a moderate interaction for method and gender . The present between method and gender effects method showed significant differences (Scheffé post hoc) with predictive equations from cadaveric measurements [2], [4] and MRI [17]. Concerning the transverse moments of inertia , ANOVA tests showed a significant effect for method , no significant effect for gender and a moderate

1760


Fig. 4. BSP obtained by the present method and estimated from the predictive equations: significant differences between the predictive equation and the present method (3 3 3p < 0:001; 3 3 p < 0:01; 3p < 0:05). (a) Segment mass M . (b) Transverse position of center of mass C . (c) Frontal moment of inertia I . Sagittal moment of inertia I .

interaction between method and gender effects . The present method showed significant differences (Scheffé post hoc) with predictive equations from cadaveric measurements [2] and MRI [17]. IV. DISCUSSION Generally, BSP are obtained from predictive equations using body mass and segment length [3], [4], [9], [11]. Significant differences between predictive equations have been reported in the literature [9], [14] and were also found in the present study. Indeed, a significant effect for method was found for all BSP ( , , , , and ). Also, a significant effect for gender was found for all BSP except . Interactions between method and gender effects were found although barely significant. This might be due to the fact that three predictive equations related to males were also used for females.

Actually, although there is a lack of study on diverse populations, predictive equations should not be used outside the population on which they are based [15], [16]. Additionally, predictive equations from various studies cannot be combined due to differences in segment boundary definitions, especially for the hip [2]–[4], [6]–[9], [19], [21], [33], [40]–[42] and, due to the measurement procedures themselves. For example, possible differences between living and cadaver tissues have been mentioned [15], [16]. That is why, significant differences were found between the present study and the predictive equations based on cadaveric measurements: the measurement procedures, segment boundaries and populations were dissimilar. As well, significant differences were found between the present study and the predictive equations based MRI: the measurement procedures were parallel, but the segment boundaries and populations were dissimilar.


Fig. 4 (Continued.). BSP obtained by the present method and estimated from the predictive equations: significant differences between the predictive equation and the present method (3 3 3p < 0:001; 3 3 p < 0:01; 3p < 0:05). (e) Transverse moment of inertia I .

On the opposite, fair agreements were found between the present study and de Leva’s and Durkin et al.’s predictive equations: although dissimilar measurement procedures, segment boundaries, and populations were comparable. These comparison mean indirect validation of the personalized BSP obtained in the present study. This indirect validation was also used in all other studies on personalized BSP, [9], [10], [17]–[21]. Direct validation could be possible [43] but hardly applicable. Presently, the scaling equations of de Leva [11] represents the most complete and practical set of predictive equations. These equations distinguish from gender and the frontal and sagittal moments of inertia are not assumed to be equivalent as in many other predictive equations. Besides, the segment endpoint definition deals with internal landmarks that are more appropriate for biomechanical models. Several predictive equations propose other segment endpoint definitions [3], [6]–[9], [44] which are not as much appropriate. The present study follows a “Body Link” definition [4], [33], [34] (i.e., form the proximal joint center to the distal joint center) that is a key point in ergonomics and dynamics assessments. As well, the present study follows anatomical recommendations [33] for the segment boundary definitions and follows the recommendations of the International Society of Biomechanics for the segment coordinate system [36]. In this segment coordinate system, the present study shows a center of mass slightly medial and posterior and a nonprincipal inertia tensor. This was also directly measured on cadaver by Chandler et al. [2]. The influence of these complete 3-D BSP should be investigated in further studies. Still, as no predictive equations provide these complete 3-D BSP, personalized computation may be useful. As well, personalized BSP are useful to study specific population when predictive equations are hardly available (females, adolescents, children) or inexistent (obeses, hemiplegics). However, the methods proposed in the literature for personalized computation [9], [10],

1761

[17]–[21] present some limitations such as 2-D only measurements, external only measurements, acquisition of multiple tomographic images, cost and radiation considerations. For example, the effective dose for thorax CT with 1-mm cuts was reported to 10 mGy [45]. As an alternative, the biplanar radiographic method is a recent 3-D imaging method that does not require multiple tomographic images. The biplanar radiographic method is already used in orthopedics for the personalized 3-D reconstructions of skeleton [22]–[26]. The validation of the biplanar radiographic method has been widely studied [26]–[32]: the accuracy was reported to 1 mm for vertebrae, pelvis, and femur. Biplanar radiographs could be obtained on 1) EOS™ or equivalent system and 2) on standard equipments using specific calibration and devices [46], [47]. However, using EOS™ system, the radiographs are low-dose. For example, the effective dose has been evaluated to 0.01 mGy for biplanar radiographs of the thorax [45]. In the present study, the external shape is considered in addition to the skeleton so that average bone tissue and soft tissue densities are used for the BSP computation. This could be paralleled with the MRI computation proposed by Cheng et al. [17]. However, the internal and external volume computed by Cheng et al. were rough frustrums between 20-mm MRI cuts. Besides, the bone tissue density used in the present study was different from 1.705 cm used by Cheng et al. [17]. Cheng et al. used this density unvaryingly for the whole body and they reported that this value may have caused overestimation for some segments. Here, the bone tissue density was 1.420 cm corresponding specifically to the femur [38] and the soft tissue density was differentiated for males and females. As the soft tissue density represents both muscle and fat, it may be inadequate for more corpulent subjects, especially when evaluating the upper leg segment of females. This could explain the larger predictive equation for segment mass found with the present method. Moreover, for the center of mass and inertia tensor, the computation may also be affected by the nonuniform density along the segment [18], [48], [49]. Nevertheless, one may consider that using both bone tissue and soft tissue is more rigorous than a simple segment density as proposed in mathematical models [41], [42], stereo-photogrammetry [1] [5]–[7], and video-based methods [20], [21]. In order to assess not only personalized volumes but also personalized densities, the potentialities of multienergy with biplanar low-dose radiography have already been investigated [26]. Finally, the use of low-dose radiography allow to expect future studies on a large number of subjects in order to establish extended data on several specific populations. V. CONCLUSION Personalized body segment parameters were determined using biplanar low-dose radiography. This method allows to measure both internal and external surfaces without requiring multiple tomographic images. This method allows to follow anatomical considerations for the definitions of segment boundaries and segment coordinate system. The parameters obtained for sixteen upper leg segments were consistent with the literature: fair agreements were found between the present study and

1762


predictive equations based on other medical imaging techniques ( -ray and DEXA). Personalized body segment parameters from biplanar low-dose radiography will allow appropriate clinical and biomechanical analysis to be performed in different fields such as ergonomics, orthopedics and rehabilitation with different populations. ACKNOWLEDGMENT The authors would like to thank the team from the Laboratoire de Biomécanique specifically S. Laporte and B. Aubert for their help in developing the 3-D reconstruction method of thigh and S. Bertrand and I. Sudhoff for their help in acquiring the X-rays. Also, the authors would like to thank F. Cheriet from the Laboratoire d’imagerie et de vision 4D (Ecole Polytechnique, Montreal, QC Canada) for giving access to video scan. REFERENCES [1] T. R. Ackland, B. A. Blanksby, and J. Bloomfield, “Inertial characteristics of adolescent male body segments,” J. Biomech., vol. 21, pp. 319–27, 1988. [2] R. F. Chandler, C. E. Clauser, J. T. McConville, H. M. Reynolds, and J. W. Young, “Investigation of Inertial Properties of the Human Body,” Aerospace Medical Research Laboratory, Wright-Patterson Air Force Base, Dayton, OH, Tech. Rep. AMRL-74-137, 1975. [3] C. E. Clauser, J. T. McConville, and J. W. Young, “Weight, Volume, and Center of Mass of Segments of the Human Body,” Aerospace Medical Research Laboratory, Wright-Patterson Air Force Base, Dayton, OH, TechRep. AMRL-TR-69-70, 1969. [4] W. T. Dempster, “Space Requirements for the Seated Operator,” Wright Air Development Center, Wright-Patterson Air Force Base, Dayton, OH, WADC Tech. Rep. TR-55-159, 1955. [5] R. K. Jenson, “Body segment mass, radius and radius of gyration proportions of children,” J. Biomech., vol. 19, pp. 359–68, 1986. [6] J. T. McConville, T. D. Churchill, I. Kaleps, C. E. Clauser, and J. Cuzzi, “Anthropometric Relationships of Body and Body Segment Moments of Inertia,” Aerospace Medical Research Laboratory, Wright-Patterson Air Force Base, Dayton, OH, Tech. Rep. AFAMRL-TR-80-119, 1980. [7] J. W. Young, R. F. Chandler, C. C. Snow, K. M. Robinette, G. F. Zehner, and M. S. Lofberg, “Anthropometric and Mass Distribution Characteristics of the Adults Female,” FAA Civil Aeromedical Institute, Oklaoma City, OK, Tech. Rep. FA-AM-83-16, 1983. [8] V. M. Zatsiorsky, V. N. Seluyanov, and L. G. Chugunova, “Methods of determining mass-inertial characteristics of human body segments,” in Contemporary Problems of Biomechanics, S. A. Regirer, Ed. Moscow: Mir, 1990, pp. 272–91. [9] J. L. Durkin and J. J. Dowling, “Analysis of body segment parameter differences between four human populations and the estimation errors of four popular mathematical models,” J. Biomech. Eng., vol. 125, pp. 515–22, 2003. [10] K. J. Ganley and C. M. Powers, “Anthropometric parameters in children: A comparison of values obtained from dual energy x-ray absorptiometry and cadaver-based estimates,” Gait Posture, vol. 19, pp. 133–40, 2004. [11] P. de Leva, “Adjustments to zatsiorsky-seluyanov’s segment inertia parameters,” J. Biomech., vol. 29, pp. 1223–30, 1996. [12] R. N. Hinrichs, “Regression equations to predict segmental moments of inertia from anthropometric measurements: An extension of the data of chandler et al. (1975),” J. Biomech., vol. 18, pp. 621–4, 1985. [13] M. R. Yeadon and M. Morlock, “The appropriate use of regression equations for the estimation of segmental inertia parameters,” J. Biomech., vol. 22, pp. 683–9, 1989. [14] D. J. Pearsall and P. A. Costigan, “The effect of segment parameter error on gait analysis results,” Gait Posture, vol. 9, pp. 173–83, 1999. [15] J. G. Reid and R. K. Jensen, “Human body segment inertia parameters: A survey and status report,” Exer. Sport Sci. Rev., vol. 18, pp. 225–41, 1990. [16] D. J. Pearsall and J. G. Reid, “The study of human body segment parameters in biomechanics. An historical review and current status report,” Sports Med., vol. 18, pp. 126–40, 1994. [17] C. K. Cheng, H. H. Chen, C. S. Chen, C. L. Chen, and C. Y. Chen, “Segment inertial properties of Chinese adults determined from magnetic resonance imaging,” Clin. Biomech., vol. 15, pp. 559–66, 2000.

[18] M. Mungiole and P. E. Martin, “Estimating segment inertial properties: Comparison of magnetic resonance imaging with existing methods,” J. Biomech., vol. 23, pp. 1039–46, 1990. [19] D. J. Pearsall, J. G. Reid, and L. A. Livingston, “Segmental inertial parameters of the human trunk as determined from computed tomography,” Ann. Biomed. Eng., vol. 24, pp. 198–210, 1996. [20] O. Sarfaty and Z. Ladin, “A video-based system for the estimation of the inertial properties of body segments,” J. Biomech., vol. 26, pp. 1011–6, 1993. [21] J. Norton, N. Donaldson, and L. Dekker, “3D whole body scanning to determine mass properties of legs,” J. Biomech., vol. 35, pp. 81–6, 2002. [22] J. P. Steib, R. Dumas, D. Mitton, and W. Skalli, “Surgical correction of scoliosis by in situ contouring: A detorsion analysis,” Spine, vol. 29, pp. 193–9, 2004. [23] R. Dumas, J. P. Steib, D. Mitton, F. Lavaste, and W. Skalli, “Three-dimensional quantitative segmental analysis of scoliosis corrected by the in situ contouring technique,” Spine, vol. 28, pp. 1158–62, 2003. [24] N. Gangnet, V. Pomero, R. Dumas, W. Skalli, and J. M. Vital, “Variability of the spine and pelvis location with respect to the gravity line: A three-dimensional stereoradiographic study using a force platform,” Surg. Radiol. Anat., vol. 25, pp. 424–33, 2003. [25] L. Nodé-Langlois, “Analyses tridimensionnelles des déviations angulaires des axes du membre inférieur, en pré per et postopératoire,” Ph.D. thesis, Ecole Nationale Supérieure d’Arts et Métiers, Paris, France, 2003. [26] A. Le Bras, “Evaluation des potentialités du système EOS pour la caractérisation mécanique de structures osseuses: Application à l’extrémité supérieure du fémur,” Ph.D. thesis, Ecole Nationale Supérieure d’Arts et Métiers, Paris, France, 2004. [27] S. Laporte, “Reconstruction 3-D du squelette humain pour la biomécanique par radiographie biplane à dose minimale d’irradiation,” Ph.D. thesis, Ecole Nationale Supérieure d’Arts et Métiers, Paris, France, 2002. [28] S. Laporte, W. Skalli, J. A. D. Guise, F. Lavaste, and D. Mitton, “A biplanar reconstruction method based on 2-D and 3-D contours: Application to the distal femur,” Comput. Methods Biomech. Biomed. Eng., vol. 6, pp. 1–6, 2003. [29] D. Mitton, C. Landry, S. Veron, W. Skalli, F. Lavaste, and J. A. De Guise, “3D reconstruction method from biplanar radiography using nonstereocorresponding points and elastic deformable meshes,” Med. Biol. Eng. Comput., vol. 38, pp. 133–9, 2000. [30] A. Mitulescu, I. Semaan, J. A. de Guise, P. Leborgne, C. Adamsbaum, and W. Skalli, “Validation of the nonstereo corresponding points stereoradiographic 3-D reconstruction technique,” Med. Biol. Eng. Comput., vol. 39, pp. 152–8, 2001. [31] V. Pomero, D. Mitton, S. Laporte, J. A. d. Guise, and W. Skalli, “Fast accurate stereoradiographic 3D-reconstruction of the spine using a combined geometric and statistic model,” Clin. Biomech., vol. 19, pp. 240–7, 2004. [32] A. Le Bras, S. Laporte, D. Mitton, J. A. de Guise, and W. Skalli, “Threedimensional (3-D) detailed reconstruction of human vertebrae from lowdose digital stereoradiograph,” Eur. J. Orthop. Surg. Traumatol., vol. 13, pp. 57–62, 2003. [33] J. P. Clarys and M. J. Marfell-Jones, “Anatomical segmentation in humans and the prediction of segmental masses from intra-segmental anthropometry,” Hum. Biol., vol. 58, pp. 771–82, 1986. [34] D. B. Chaffin and G. B. J. Anderson, Occupational Biomechanics. New York: Wiley, 1991. [35] A. Cappozzo, F. Catani, U. D. Croce, and A. Leardini, “Position and orientation in space of bones during movement: Anatomical frame definition and determination,” Clin. Biomech. (Bristol, Avon), vol. 10, pp. 171–178, 1995. [36] G. Wu, S. Siegler, P. Allard, C. Kirtley, A. Leardini, D. Rosenbaum, M. Whittle, D. D. D’Lima, L. Cristofolini, H. Witte, O. Schmid, and I. Stokes, “ISB recommendation on definitions of joint coordinate system of various joints for the reporting of human joint motion—Part I: Ankle, hip, and spine. International society of biomechanics,” J. Biomech., vol. 35, pp. 543–8, 2002. [37] B. Mirtich, “Fast and accurate computation of polyhedral mass properties,” J. Graphic Tools, vol. 1, pp. 31–50, 1996. [38] D. R. White, H. Q. Woodard, and S. M. Hammond, “Average soft-tissue and bone models for use in radiation dosimetry,” Br. J. Radiol., vol. 60, pp. 907–13, 1987. [39] P. de Leva, “Joint center longitudinal positions computed from a selected subset of Chandler’s data,” J. Biomech., vol. 29, pp. 1231–3, 1996. [40] D. J. Pearsall, J. G. Reid, and R. Ross, “Inertial properties of the human trunk of males determined from magnetic resonance imaging,” Ann. Biomed. Eng., vol. 22, pp. 692–706, 1994.


[41] E. P. J. Hanavan, “A Mathematical Model of the Human Body,” Aerospace Medical Research Laboratory, Wright-Patterson Air Force Base, Dayton, OH, Technical Report AMRL-TR-64-102, 1964. [42] H. Hatze, “A mathematical model for the computational determination of parameter values of anthropomorphic segments,” J. Biomech., vol. 13, pp. 833–43, 1980. [43] T. C. Pataky, V. M. Zatsiorsky, and J. H. Challis, “A simple method to determine body segment masses in vivo: Reliability, accuracy and sensitivity analysis,” Clin. Biomech., vol. 18, pp. 364–8, 2003. [44] R. K. Jensen, “Estimation of the biomechanical properties of three body types using a photogrammetric method,” J. Biomech., vol. 11, pp. 349–58, 1978. [45] S. Ferey, I. Dorion, M. Meynadier, L. Miladi, C. Maccia, D. Mitton, and G. Kalifa, “Résultats dosimétriques du système numérique EOS: de l’attitude alara à une réelle application pratique,” J. Radiol., vol. 85, no. 9, p. 1459, 2004. [46] R. Dumas, D. Mitton, S. Laporte, J. Dubousset, J. P. Steib, F. Lavaste, and W. Skalli, “Explicit calibration method and specific device designed for stereoradiography,” J. Biomech., vol. 36, pp. 827–34, 2003. [47] F. Cheriet and J. Meunier, “Self-calibration of a biplane X-ray imaging system for an optimal three dimensional reconstruction,” Comput. Med. Imag. Graph, vol. 23, pp. 133–41, 1999. [48] T. R. Ackland, P. W. Henson, and D. A. Bailey, “The uniform density assumption: Its effect upon the estimation of body segment inertial parameters,” Int. J. Sports Biomech., vol. 4, pp. 146–55, 1988. [49] C. Wei and R. K. Jensen, “The application of segment axial density profiles to a human body inertia model,” J. Biomech., vol. 28, pp. 103–8, 1995.

Raphaël Dumas received the B.Sc. degree in mechanical engineering from the Institut National des Sciences Appliquées, Lyon, France, in 1998 and the Ph.D. degree in biomechanics from the Ecole Nationale Supérieure d’Arts et Métiers, Paris, France, in 2002. He joined the Laboratoire de recherche en imagerie et orthopédie for his postdoctoral research from 2003 to 2004. He his currently Assistant Professor at the Université Claude Bernard and member of the Laboratoire de Biomécanique et Modélisation Humaine. His research interests are related to 3-D personalized human modeling for biomechanics.

Rachid Aissaoui (M’98) received the B.Sc. degree in electrical engineering from the University of Science and Technology of Oran, Oran, Algeria, in 1985 and the Ph.D. degree in biomechanics from the Université Joseph Fourier, Grenoble, France, in 1990. He joined the Clinical Research Institute of Montreal, QC, Canada, as head of the Engineering Rehabilitation team in 1991. He was responsible for the gait laboratory at the Sainte-Justine Hospital Research Center, Montreal, from 1992 to 1995. He worked as a Researcher at the NSERC Industrial Research Chair on Wheelchair Seating Aids, Montreal, from 1996–2001. He is currently Full Professor with the department of Génie de la Production Automatisée at Ecole de Technologie Supérieure and member of the Laboratoire de recherche en imagerie et orthopédie, Montreal. His research interests are related to the 3-D modeling of human locomotion, the development of tools for seating posture evaluation, and the dynamics of wheelchair propulsion. Dr. Aissaoui is a member of the CRIR, REPAR, and the IEEE-EMB Society.

1763

David Mitton received the M.Sc. degree from the Université Claude Bernard, Lyon, France, in 1994. He received the Ph.D. degree in biomechanics from the Institut National des Sciences Appliquées, Lyon, in 1997. He is a Biomedical Engineer. He joined the Laboratoire de Biomécanique in 1998, where he has been an Associate Professor since 2003. His research is oriented toward mechanical behavior of musculoskeletal system with the aim to define geometric modeling of the skeleton and to characterize mechanical behavior of biological tissues. Dr. Mitton is a member of the Société de Biomécanique (international french speaking Society of Biomechanics) and of the council of this Society.

Wafa Skalli received the B.Sc. degree in 1980 and the Ph.D. degree in 1983 from the Ecole Nationale Supérieure d’Arts et Métiers, Paris, France. She joined the Laboratoire de Biomécanique in 1988, where she is currently Full Professor and Assistant Director, in charge of research team “osteo-articular biomechanics and clinical research.” Her research interests are to better understand mechanisms of degradation, to improve implants and to provide quantitative tools for clinical assistance in diagnosis, follow up or surgical planning. She is particularly involved in personalized geometric and mechanical modeling, with a strong link to experimental and clinical approach, and in close collaboration with clinical teams of various hospitals.

Jacques A. De Guise received the B.Sc. degree in electrical engineering and the Ph.D. degree in biomedical engineering from École Polytechnique of Montreal, Montreal, QC, Canada, in 1977 and 1984, respectively. He was a Natural Sciences and Engineering Research Council (NSERC) Postdoctoral Scholar at the Computer Vision and Robotics Laboratory of McGill University (Canada) from 1984 to 1986. He was a NSERC Researcher Fellow at the Institut de génie biomédical of the University of Montreal from 1986 to 1990. He is currently Full Professor at the Automated Production Department of the École de technologie supérieure of Montreal and Director of the Laboratoire de recherche en imagerie et orthopédie (LIO) of the University of Montreal Hospital Research Centre. He is Chair holder of the Canada Research Chair in 3-D imaging and biomedical engineering. His current research interests are 3-D medical imaging, 3-D modeling of the musculoskeletal and vascular systems, and computer assisted surgery.