Elbow-flexion force estimation during arm posture dynamically changing between pronation and supination. (10th October 2019)

• Record Type: Journal Article
• Authors: Hu, Ruochen; Chen, Xiang; Huang, Chengjun; Cao, Shuai; Zhang, Xu; Chen, Xun
• Is Part Of: Journal of neural engineering. Volume 16:Number 6(2019:Dec.)
• Publication Date: 2019-10-10
• Subjects: dynamic voluntary contraction task -- isometric elbow flexion -- Force estimation -- High-density surface EMG -- NMF algorithm; Neurosciences -- Periodicals; Biomedical engineering -- Periodicals; 612.8
• DOI: 10.1088/1741-2552/ab2e18
• Languages: English
• ISSNs: 1741-2560

Abstract

Abstract: Objective . In this study, we proposed a dynamic voluntary contraction force estimation framework and implemented it for elbow-flexion force estimation during arm posture dynamically changing between pronation and supination. Approach . High-density surface electromyography (HD-sEMG) from biceps brachii and brachialis muscles and the elbow-flexion force were measured synchronously. The simplified Hill model was adopted to establish the relation between HD-sEMG and the elbow-flexion force. In the training process of the force estimation model, HD-sEMG data from static isometric elbow flexion tasks in two force modes (staircase and sinusoidal) and two postures (supination or pronation) were used. The nonnegative matrix factorization (NMF) algorithm was adopted to decompose HD-sEMG into activation patterns and the corresponding time-varying coefficient vectors. The major activation pattern was used to select the appropriate sEMG channels for extracting the input signal of the force estimation model. In the testing phase, the elbow-flexion force estimation during arm posture dynamically changing between pronation and supination was conducted. HD-sEMG was also decomposed using NMF algorithm. In this case, the concept of the major activation pattern is no longer applicable because the activation areas of biceps brachii and brachialis vary with the change of arm posture. An improved channel selection scheme based on the ratio of activation intensities was presented to extract the input signal of the well-trained simplified Hill model . Main results . (1) the improved channel selection scheme could locate effectively the primary muscle activation areas related to the change of arm posture; (2) the input signal extraction method based on the ratio of activation intensities obtained the best force estimation performance compared to the method based on all channels of HD-sEMG array and the method based on major activation pattern; (3) the force estimation performance was better when the simplified Hill model was calibrated with the sinusoidal force data rather than staircase data. Significance . This research provides an effective solution to realize muscle force estimation during a dynamic voluntary contraction task. It can be further extended to the research fields of biomechanics, sports, and rehabilitation medicine.