三次元筋骨格モデルにおける下肢筋走行の再構築と精度検証

抄録

To analyze for osteoarthritis patients, it is considered clinically useful to predict biological force by the lower extremity musculoskeletal model. However, most of them are constructed by straight muscles (Via-Point muscle), and the number of lower extremity models that reproduce actual muscle path is few. In this study, we developed the lower extremity musculoskeletal model which reproduced the actual muscle path and validated the accuracy by using motion analysis data. First, we constructed a Wrapping muscle that reproduces the curvilinear shape of the muscle by putting the Via-Point muscle along multiple ellipsoids provided by AnyBody Modeling System (AMS) Via-Point muscle. Secondly, we conducted the inverse dynamics analysis using motion analysis data to validate the accuracy of the developed model. For the verification, we used the walking data of Total Hip Arthroplasty patient and the hip contact force (HCF). We conducted the inverse dynamics analysis to calculate the HCF and muscle activations. We calculated the mean squared error (RMSE) of the predicted value and the actual value for the HCF calculated by resulting the three axial direction components. As well as, we also calculated RMSE of the default model to compare the developed model. As a result, the RMSE of measured total HCF and predicted total HCF was 0.208 * BW for measured total HCF. In addition, the RMSE of HCF Predicted calculated with the default model provided by AMS is 0.350 * BW, which showed to reduce the error by 40%. Based on the results of this study, it was possible to develop a model with higher accuracy than in the past by reconsidering the origin and insertion of the muscle path into a shape closer to that of the actual body.