ABSTRACT
Patient-specific modeling is a vital component in the translation of computational multibody dynamics into clinical practice. Recent research has focused on ways to derive such models from medical imaging, but the process is usually time consuming and sensitive to operator skill. Here, we present methods to derive kinematic and inertial properties of body segments from MRI images, and condense them into a dynamically consistent patient-specific multibody model (PSM). We develop a semi-automated tool chain to classify bone, muscle and fat in the lower body and use optimization and geometrical methods to derive personalized bone meshes and segment inertial properties. The tool chain is applied to investigate the gait of a 12-yr old female with bone deformities. The patient-specific results are compared to those arising from generic scaled models with parameters based on regression equations. We found several kinematic and inertial differences between the two models, and overall the PSM resulted in markedly smaller angular and force residuals. The PSM was able to capture vital aspects of this patient׳s gait in the transverse plane that were overlooked by the generic model. These results are relevant to the use of multibody dynamics in the planning of surgical interventions, and form the basis for developing efficient and automatic methods to create patient-specific models.
