Multimodal Medical Imaging Fusion for Patient Specific Musculoskeletal Modeling of the Lumbar Spine System in Functional Posture
Tien Tuan Dao,
Marie Christine Ho Ba Tho
Current musculoskeletal modeling does not take the functionally initial standing posture of the lumbar spine system into consideration. This may lead to inaccurate simulation outcomes for clinical decision support. This present study aimed to develop a fusion process from multimodal medical images for developing, simulating, and assessing patient specific musculoskeletal model of the lumbar spine system in functional posture. Computed tomography (CT) and X-rays were acquired on a patient. A 3D/2D matching procedure was developed using feature-based approach to transform CT-based 3D geometries into standing posture. Then, a musculoskeletal model was developed including 7 segments, 18-DOF (degrees of freedom) and 126 muscle fascicles. Model evaluation was performed using X-ray data and curvature metric. Muscle force deviations were quantified for the comparison between image-based model and generic-scaled model as well as for the sensitivity of psoas major attachment point definition. Curvatures of the simulation outcomes fall within the range of X-ray data for extension posture. Difference was noted for flexed posture. The use of generic model leads to a deviation of 66% of the muscle forces according to the image-based model. A maximal relative deviation of 4% of the estimated psoas major fascicle force was found for the analysis of the sensitivity of muscle attachment point. This study proposed a useful data fusion process from multimodal medical images to create patient specific lumbar spine model in correct initial posture leading to perform accurate dynamic simulations. Moreover, this study suggested that image-based model needs to be developed for clinical decision making related to spinal disorders.