Combining ultrafast ultrasound and high-density EMG to assess electromechanical muscle properties.

Assessing muscle properties is relevant in several contexts, from the study of muscle physiology/pathophysiology to the assessment of rehabilitation outcomes. Muscle activation and the resulting tissue displacement are traditionally studied with surface electromyograms (sEMGs) and ultrasound (US) images respectively. By recording sEMG signals from several locations above the skin surface (high-density sEMG - HDEMG), it is possible to study the spatial distribution of EMG activity over the muscle and to decompose the interferential signal into its constituent trains of motor unit action potentials (MUAPs). High frame rate US imaging allows to reveal fine spatial details of tissue displacement with high temporal resolution (up to 5,000 fps), enabling in vivo estimation of the mechanical muscle properties from tissue velocity fields. The aim of this dissertation is to combine ultrafast ultrasound and HDEMG to assess the electromechanical properties of single motor units (MU). A method for the identification of single MU territories during low-level isometric muscle contractions in fusiform muscles is proposed. The method exploits two main pieces of information: the spatio-temporal evolution of tissue velocity, obtained from B-mode ultrasound videos, and the instants of MUs activation, obtained from HDEMG decomposition. Specifically, we tested two approaches: in the first one we applied spike triggered averaging (STA) on the tissue velocity sequence (TVS) based on the times of occurrence of individual MUAPs extracted from HDEMG decomposition; the second was based on the correlation between the decomposed MU firing patterns and the spatio-temporal independent components (STIC) of the TVS. Before testing the two approaches on experimental data, the algorithms were refined and validated on simulated data. To this end, an EMG simulator was adapted to include single MU mechanical responses thus providing a unified description of electromechanical muscle activation. The effect of the following factors was tested for both algorithms: signal length, contraction level, MU characteristics (firing properties, size distributions) and mechanical properties of the muscle tissue (passive force transmission in the transversal muscle plane). The performances of the two algorithms were quantified through the comparison between the simulated and the identified characteristics of MU territories (size and location) and the temporal evolution of the mechanical twitch. Experimental signals were collected from one subject during low level (2 % and 5% MVC) isometric contractions of the biceps brachii. For both algorithms, we observed a match between the simulated and identified MU territories, as well as a match between the mediolateral positions of MU territories and the barycentre of the corresponding MUAP distributions. However, while the performance of STA algorithm decayed rapidly with the increase of the contraction level (i.e. the number of active MU), those of the method based on the STIC did not change up to 20 %MVC. When applied to experimental data only this latter method was able to show localized mechanical responses, whose position matched with the surface representation of MU potential. Although preliminary, the results of this Thesis suggest that the electromechanical characteristics of individual MUs can be non-invasively assessed at low contraction levels through the integration of high frame rate ultrasound and high density EMG.