Isometric Plantarflexion Moment Prediction Based on a Compartment-specific HD-sEMG-driven Musculoskeletal Model.

OBJECTIVE: electromyogram (EMG)-driven musculoskeletal models have been widely used to investigate human movements while existing EMG-driven models commonly neglect regional heterogeneity in anatomy and activation within a skeletal muscle. To consider neuromuscular compartment anatomy and activation, a subject- and compartment-specific EMG-driven model was developed for isometric plantarflexion moment prediction.

METHODS: the model was hill-type consisting of gastrocnemius medialis, gastrocnemius lateralis, and soleus around the ankle joint, and each muscle was discretised into four compartments. The moment arms of each compartment were determined using magnetic resonance imaging and the compartment activation was calculated based on high-density surface EMG signals. And the hill-type compartment parameters were tuned in a calibration process. The developed compartment-specific model and a generic EMG-driven model were examined by comparing their predicted net ankle moments with measurements obtained while subjects performed isometric plantarflexion tasks at different contraction levels.

RESULTS: compared to the generic EMG-driven model, the isometric plantarflexion moment prediction using the compartment-specific model was more accurate at all contraction levels, with the average prediction error decreasing from average 13.81% to 10.11%. The contraction of each compartment was found to be generally non-uniform at all contraction levels.

CONCLUSION: the developed compartment-specific model enabled accurate prediction of isometric plantarflexion moment and the simulation of non-uniform muscular contraction, which is more physiologically appropriate than the existing EMG-driven models.

SIGNIFICANCE: the proposed compartment-specific formulation opens new perspectives for subject-specific musculoskeletal modelling, which has great potential in understanding regional characteristics of the neuromuscular activities.