A Subject-Specific EMG-Driven Musculoskeletal Model for the Estimation of Moments in Ankle Plantar-Dorsiflexion Movement
In traditional rehabilitation process, ankle movement ability is only qualitatively estimated by its motion performance, however, its movement is actually achieved by the forces acting on the joints produced by muscles contraction. In this paper, the musculoskeletal model is introduced to provide a more physiologic method for quantitative muscle forces and muscle moments estimation during rehabilitation. This paper focuses on the modeling method of musculoskeletal model using electromyography (EMG) and angle signals for ankle plantar-dorsiflexion (P-DF) which is very important in gait rehabilitation and foot prosthesis control. Due to the skeletal morphology differences among people, a subject-specific geometry model is proposed to realize the estimation of muscle lengths and muscle contraction force arms. Based on the principle of forward and inverse dynamics, difference evolutionary (DE) algorithm is used to adjust individual parameters of the whole model, realizing subject-specific parameters optimization. Results from five healthy subjects show the inverse dynamics joint moments are well predicted with an average correlation coefficient of 94.21% and the normalized RMSE of 12.17%. The proposed model provides a good way to estimate muscle moments during movement tasks.
KeywordsEMG signals Musculoskeletal model Ankle plantar-dorsiflexion Joint moment
Research supported by The Excellent Dissertation Cultivation Funds of Wuhan University of Technology with No. 2016-YS-062 and National Natural Science Foundation of China under grants Nos. 51475342 and 61401318.
- 1.Zhang, T.: Stroke rehabilitation in China (2011 edition). Chin. J. Rehabil. Theor. Pract. 18(4), 301–318 (2012). (in Chinese)Google Scholar
- 5.Meng, W., Ding, B., Zhou, Z., et al.: An EMG-based force prediction and control approach for robot-assisted lower limb rehabilitation. In: 45th IEEE International Conference on Systems, Man, and Cybernetics, pp. 2198–2203. Institute of Electrical and Electronics Engineers Inc., San Diego (2014)Google Scholar
- 8.Kurt, M., Karin, G., Buchanan, T.: A real-time EMG-driven musculoskeletal model of the ankle. Multibody Sys. Dyn. 28(1–2), 169–180 (2012)Google Scholar
- 9.Zhang, M., Meng, W., Davies, T., et al.: A robot-driven computational model for estimating passive ankle torque with subject-specific adaptation. IEEE Trans. Biomed. Eng. 63(4), 814–821 (2016)Google Scholar
- 13.Zheng, R., Liu, T., Kyoko, S., et al.: In vivo estimation of dynamic muscle-tendon moment arm length using a wearable sensor system. In: 12th IEEE/ASME International Conference on Advanced Intelligent Mechatronics, pp. 647–652. Institute of Electrical and Electronics Engineers Inc., Xi’an (2008)Google Scholar