Neuromuscular characterisation in Cerebral Palsy using hybrid Hill-type models on isometric contractions. (1st December 2018)

• Record Type: Journal Article
• Authors: Wiedemann, L.G.; Jayaneththi, V.R.; Kimpton, J.; Chan, A.; Müller, M.A.; Hogan, A.; Lim, E.; Wilson, N.C.; McDaid, A.J.
• Is Part Of: Computers in biology and medicine. Volume 103(2018)
• Page Start: 269
• Page End: 276
• Publication Date: 2018-12-01
• Subjects: Musculoskeletal modelling -- Hill-type muscle model -- Cerebral palsy -- Neuromuscular characterisation -- Electromyography -- HD-EMG -- Isometric contraction -- Force ramp; Medicine -- Data processing -- Periodicals; Biology -- Data processing -- Periodicals; 610.285
• DOI: 10.1016/j.compbiomed.2018.10.027
• Languages: English
• ISSNs: 0010-4825

Abstract

Abstract: Background: Muscles of individuals with Cerebral Palsy (CP) undergo structural changes over their lifespan including an increase in muscle stiffness, decreased strength and coordination. Being able to identify these changes non-invasively would be beneficial to improve understanding of CP and assess therapy effectiveness over time. This study aims to adapt an existing EMG-driven Hill-type muscle model for neuromuscular characterisation during isometric contractions of the elbow joint. Methods: Participants with (n = 2) and without CP (n = 8) performed isometric force ramps with contraction levels ranging between 15 and 70% of their maximum torque. During these contractions, high-density EMG data were collected from the M. Biceps and Triceps brachii with 64 electrodes on each muscle. The EMG-driven Hill-type muscle model was used to predict torques around the elbow joint, and muscle characterisation was performed by applying a genetic algorithm that tuned individuals' parameters to reduce the RMS error between observed and predicted torque data. Results: Observed torques could be predicted accurately with an overall mean error of 1.24Nm ± 0.53Nm when modelling individual force ramps. The first four parameters of the model could be identified relatively reliably across different experimental protocols with a full-scale variation of below 20%. Conclusion: An HD-EMG muscle modelling approach to evaluating neuromuscular properties in participants with and without CP has been presented. This pilot study confirms the feasibility of the experimental protocol and demonstrates some parameters can be identified robustly using the isometric contraction force ramps. Highlights: A non-invasive hybrid model to characterise muscle parameters was developed. Subject-specific muscle parameters of subjects with Cerebral Palsy were evaluated. PCA processing of HD-EMG resulted in improved model accuracy. A cross-validation study revealed good model generalisability. Low assessment variation was present for four out of ten model parameters.