• Luca Mesin.
• Mathematical Biology and Physiology, Dipartimento di Elettronica e Telecomunicazioni, Politecnico di Torino, Corso Duca degli Abruzzi 24, 10129 Turin, Italy. Electronic address: luca.mesin@polito.it.
• Comput. Biol. Med. 2015 Feb 1; 57: 8-19.

PurposeDeveloping a real time method to estimate generation, extinction and propagation of muscle fibre action potentials from bi-dimensional and high density surface electromyogram (EMG).MethodA multi-frame generalization of an optical flow technique including a source term is considered. A model describing generation, extinction and propagation of action potentials is fit to epochs of surface EMG.ResultsThe algorithm is tested on simulations of high density surface EMG (inter-electrode distance equal to 5mm) from finite length fibres generated using a multi-layer volume conductor model. The flow and source term estimated from interference EMG reflect the anatomy of the muscle, i.e. the direction of the fibres (2° of average estimation error) and the positions of innervation zone and tendons under the electrode grid (mean errors of about 1 and 2mm, respectively). The global conduction velocity of the action potentials from motor units under the detection system is also obtained from the estimated flow. The processing time is about 1 ms per channel for an epoch of EMG of duration 150 ms.ConclusionsA new real time image processing algorithm is proposed to investigate muscle anatomy and activity. Potential applications are proposed in prosthesis control, automatic detection of optimal channels for EMG index extraction and biofeedback.Copyright © 2014 Elsevier Ltd. All rights reserved.

17 keywords...

hide…