Real time identification of active regions in muscles from high density surface electromyogram

PURPOSE Developing a real time method for the localization of muscle activity regions from high density surface electromyogram (EMG). METHOD The inverse problem of source localization is solved by a regularized technique applied to an over-determined problem searching for the least mean squares approximation of the recorded signal with a linear combination of a set of basis waveforms (subject specific). RESULTS The method, tested on simulations, provides accurate estimates of the mean location of the sources (in ideal conditions, it has about 1 mm of mean error in locating the depth, negligible error in locating the transverse location of the active region). For reasonably small perturbations, it is stable to possible detection problems (e.g., misalignment between the electrodes and the fibres, noise), inaccurate knowledge of the anatomical and physical properties of the investigated tissues (e.g., tissue thickness, location of IZ, fibre length, tissue conductivity) with mean estimation errors of about 1.5-2.8 mm. CONCLUSIONS An innovative algorithm is proposed for the non-invasive localization of the active regions of a muscle. It is real time and opens potential future applications for prosthesis control and biofeedback.