Figure 1

Overview of the proposed algorithm. (a) HDsEMG signals of a grid of electrodes. (b) Decomposed MU firing pattern from HDsEMG signals. (c) Generation of the train of MU displacement velocities through the convolution of the MU firing pattern with the synthetic velocity profile. (d) Ultrafast US B-mode sequence with 3 examples of ROI used in the analysis. (e) Tissue Velocity Sequence estimated using 2D autocorrelation approach. (f) Output of the Singular Value Decomposition and spatio-temporal Independent Component Analysis of the three example ROIs with 50 components (comp.) each composed by a time course (temporal component) and a correspondent image (spatial component). (g) Integration between EMG and US variables: cross-correlations between all the temporal components of the tissue velocity sequence of the ROIs and the synthetic train of MU displacement velocities, and example outcome for a single MU. (1) Map of maximum correlation coefficient of all the ROIs (19 × 19). The algorithm extracted the component with the maximum correlation within ± 20 ms time lag for each ROI. The selected ROIs (cluster) are highlighted with black dots. (2) heatmap of the MU spatial representation in US images extracted summing the spatial components of the most correlated ROIs in the coefficient map (black dots); (3) spike triggered averaged velocity profile (black solid line) and its standard deviation (grey band).