Fig. 1

Framework for MU identification in ultrafast ultrasound (UUS) image sequences was composed of four stages. A The first stage; data acquisition. Collecting synchronized UUS and concentric needle electromyography (EMG) measurements on the biceps brachii under low force voluntary isometric contractions. B The second stage; calculating tissue velocities (based on the UUS radiofrequency signals). C The third stage; data decomposition. We inserted each region-of-interest (ROI, 25 in total) into four different decomposition algorithms (see Table 1) to extract 25 spatiotemporal components. D The fourth and final stage; post-processing. We selected one optimal component out of 625 (25 components in each of the 25 ROIs) based on its distance to the needle tip (< 10 mm) and maximal agreement to MU firing rate in terms of RoA. The selected components’ features are then compared between the different decomposition algorithms