1
S346 Presentation 0078, Body Segment Inertial Parameters. 16:45, Room 105 A PATIENT SPECIFIC MODEL FOR GAIT ANALYSIS I Südhoff 1,2 , W Skalli 1 , A Fuentes 2 , JA de Guise 2 1 Laboratoire de biomécanique, CNRS UMR 8005 - École nationale supérieure d'arts et métiers, Paris, France 2 Laboratoire de recherche en imagerie et orthopédie, Centre de recherche du CHUM et École de technologie supérieure, Montréal, Canada ; email: [email protected] INTRODUCTION In gait analysis, joint forces and moments are computed from force plate data, kinematics data and body segment parameters (BSP). BSP are estimated from predictive or scaling equations, such as de Leva’s equations [1], which are limited to the three principal moments of inertia and to the center of mass’ position along a line joining two adjacent joints. Dumas proposed a method to compute personalized BSP of the thigh thanks to stereoradiography [2]. The object of this work is to combine stereoradiography and gait analysis to set up a patient specific model of the lower limb in order to enhance the accuracy of computed kinetics and improve the clinical analysis. METHODS 10 subjects participated in the protocol, which consisted in: 1. a gait analysis using a VICON ® system, an instrumented treadmill, and a marker attachment system to limit soft tissue artefacts [3]. 2. a stereo-radiographic acquisition in a standing posture, using the low dose bi-planar X-ray EOS ® system. The subjects’ bones and skin-envelope were reconstructed using specialized software [4]. A functional reference system was defined in both the VICON and the EOS system. For five subjects, personalized BSP of the lower limb were computed by adapting the method proposed by Dumas [2] and compared to those of de Leva. Lower limb kinetics were calculated using both de Leva and personalized BSP. The impact on kinetics was assessed by computing the rms between the average curves (among approximately 20 gait cycles) of each force and moment component obtained with de Leva and personalized BSP. The maximal difference between both average curves was computed and divided by the range of value of the studied component. RESULTS AND DISCUSSION As previously observed by Dumas, the center of mass (CM) is posterior (9mm) and medial (3mm) to the long axis of the thigh (Table 1). For the shank, it is 18mm posterior to the longitudinal axis. The products of inertia are very low. The generic thigh is 6.5% heavier and the shank 4.2% lighter than the patient specific one. The principal moments of inertia (I transverse , I sagittal , I longitudinal ) differ by up to 20%, the most important difference being about the longitudinal axis (Table 2). Table 1: Personalized BSP obtained for 5 subjects Patient-specific values Thigh Shank Mass (kg) 9,3 3,1 CM longitudinal (from the knee, in mm) -234,0 189,0 CM transverse (lat-med axis, in mm) -5,8 -6,7 CM sagittal (post-ant axis, in mm) -2,7 -18,0 Izz longitudinal (g m²) 33,4 4,4 Ixx transverse (g m²) 167,2 36,2 Iyy sagittal (g m²) 163,7 36,1 Ixy (g m²) 2,8 0,2 Ixz (g m²) -9,4 -0,5 Iyz (g m²) -9,4 -0,8 Table 2: Relative difference between personalized and de Leva BSP (in%) relative difference (in %) Thigh Shank Mass -6,48 4,19 CM longitudinal position -6,99 1,1 Ixx transverse -14,59 7,85 Iyy sagittal -16,29 2,46 Izz longitudinal -16,99 -20,86 As previously observed by Ganley [5] and Pearsall [6], the impact on kinetics is relatively weak: the rms between forces and moments computed with de Leva and personalized BSP is inferior to 14N for the forces and 4Nm for the moments. The shape of the curves is not changed, only some peak values differ by up to 18% for the moments and 15% for the forces of the components’ range (Figure 1). Figure 1: Medio-lateral force at the knee and hip flexion moment with de Leva (-) and personalized (.) BSP CONCLUSIONS AND FUTURE WORK The feasibility of combining 3D internal and external gait analysis data has been assessed. Patient specific 3D and de Leva BSP differ by up to 20%. The impact on kinetics curves seems to be relatively weak; however peak values can change by up to 18% of the components’ range. The next step of this study will consist in registering the bones into the gait analysis frame to provide the joint centers’ precise position and enhance the kinetics’ accuracy. Moreover, visualizing the moving bones and expressing forces and moments in bone-specific reference systems will ease the physician’s clinical analysis of kinematic and kinetic data. REFERENCES 1. De Leva P, et al. J Biomech 29, 1223-1230, 1996 2. Dumas R, et al. IEEE Transactions on Biomedical Engineering 52, 1756-1763, 2005 3. Ganjikia S, et al. The Knee 7(4), 221-231, 2000 4. Laporte S, et al. Comput Methods Biomech Biomed Engin 6 (1), 1-6, 2003 5. Ganley K, et al Clin. Biomech 19 (1), 50-6, 2004 6. Pearsall DJ, et al Gait&Posture 9 (3), 173-83, 1999 Journal of Biomechanics 40(S2) XXI ISB Congress, Podium Sessions, Tuesday 3 July 2007

A PATIENT SPECIFIC MODEL FOR GAIT ANALYSIS

Embed Size (px)

Citation preview

Page 1: A PATIENT SPECIFIC MODEL FOR GAIT ANALYSIS

S346 Presentation 0078, Body Segment Inertial Parameters. 16:45, Room 105

A PATIENT SPECIFIC MODEL FOR GAIT ANALYSIS

I Südhoff1,2, W Skalli1, A Fuentes2, JA de Guise2

1 Laboratoire de biomécanique, CNRS UMR 8005 - École nationale supérieure d'arts et métiers, Paris, France 2 Laboratoire de recherche en imagerie et orthopédie, Centre de recherche du CHUM et École de technologie supérieure,

Montréal, Canada ; email: [email protected]

INTRODUCTION In gait analysis, joint forces and moments are computed from force plate data, kinematics data and body segment parameters (BSP). BSP are estimated from predictive or scaling equations, such as de Leva’s equations [1], which are limited to the three principal moments of inertia and to the center of mass’ position along a line joining two adjacent joints. Dumas proposed a method to compute personalized BSP of the thigh thanks to stereoradiography [2]. The object of this work is to combine stereoradiography and gait analysis to set up a patient specific model of the lower limb in order to enhance the accuracy of computed kinetics and improve the clinical analysis.

METHODS 10 subjects participated in the protocol, which consisted in: 1. a gait analysis using a VICON® system, an instrumented treadmill, and a marker attachment system to limit soft tissue artefacts [3]. 2. a stereo-radiographic acquisition in a standing posture, using the low dose bi-planar X-ray EOS® system.

The subjects’ bones and skin-envelope were reconstructed using specialized software [4]. A functional reference system was defined in both the VICON and the EOS system. For five subjects, personalized BSP of the lower limb were computed by adapting the method proposed by Dumas [2] and compared to those of de Leva. Lower limb kinetics were calculated using both de Leva and personalized BSP. The impact on kinetics was assessed by computing the rms between the average curves (among approximately 20 gait cycles) of each force and moment component obtained with de Leva and personalized BSP. The maximal difference between both average curves was computed and divided by the range of value of the studied component.

RESULTS AND DISCUSSION As previously observed by Dumas, the center of mass (CM) is posterior (9mm) and medial (3mm) to the long axis of the thigh (Table 1). For the shank, it is 18mm posterior to the longitudinal axis. The products of inertia are very low. The generic thigh is 6.5% heavier and the shank 4.2% lighter than the patient specific one. The principal moments of inertia (Itransverse, Isagittal, Ilongitudinal) differ by up to 20%, the most important difference being about the longitudinal axis (Table 2).

Table 1: Personalized BSP obtained for 5 subjectsPatient-specific values Thigh ShankMass (kg) 9,3 3,1CM longitudinal (from the knee, in mm) -234,0 189,0CM transverse (lat-med axis, in mm) -5,8 -6,7CM sagittal (post-ant axis, in mm) -2,7 -18,0Izz longitudinal (g m²) 33,4 4,4Ixx transverse (g m²) 167,2 36,2Iyy sagittal (g m²) 163,7 36,1Ixy (g m²) 2,8 0,2Ixz (g m²) -9,4 -0,5Iyz (g m²) -9,4 -0,8

Table 2: Relative difference between personalized and de Leva BSP (in%)

relative difference (in %) Thigh ShankMass -6,48 4,19CM longitudinal position -6,99 1,1Ixx transverse -14,59 7,85Iyy sagittal -16,29 2,46Izz longitudinal -16,99 -20,86

As previously observed by Ganley [5] and Pearsall [6], the impact on kinetics is relatively weak: the rms between forces and moments computed with de Leva and personalized BSP is inferior to 14N for the forces and 4Nm for the moments. The shape of the curves is not changed, only some peak values differ by up to 18% for the moments and 15% for the forces of the components’ range (Figure 1).

Figure 1: Medio-lateral force at the knee and hip flexion moment with de Leva (-) and personalized (.) BSP

CONCLUSIONS AND FUTURE WORK The feasibility of combining 3D internal and external gait analysis data has been assessed. Patient specific 3D and de Leva BSP differ by up to 20%. The impact on kinetics curves seems to be relatively weak; however peak values can change by up to 18% of the components’ range. The next step of this study will consist in registering the bones into the gait analysis frame to provide the joint centers’ precise position and enhance the kinetics’ accuracy. Moreover, visualizing the moving bones and expressing forces and moments in bone-specific reference systems will ease the physician’s clinical analysis of kinematic and kinetic data. REFERENCES 1. De Leva P, et al. J Biomech 29, 1223-1230, 1996 2. Dumas R, et al. IEEE Transactions on Biomedical

Engineering 52, 1756-1763, 2005 3. Ganjikia S, et al. The Knee 7(4), 221-231, 2000 4. Laporte S, et al. Comput Methods Biomech Biomed Engin

6 (1), 1-6, 2003 5. Ganley K, et al Clin. Biomech 19 (1), 50-6, 2004 6. Pearsall DJ, et al Gait&Posture 9 (3), 173-83, 1999

Journal of Biomechanics 40(S2) XXI ISB Congress, Podium Sessions, Tuesday 3 July 2007