Fig. 3: Factorization algorithms to assess the spatiotemporal and spatiofrequency organization of the muscle activity patterns.
a Type of multidimensional decomposition of EMG data (m) in the spatiotemporal domain: spatial decomposition, temporal decomposition, and space-by-time decomposition (from top to bottom). To assess consistent muscle modules across strides (s) the criteria of similarity was used: scalar product between basic patterns (c) for spatial decomposition, scalar product between synergies (w) for temporal decomposition, diagonality of the activation coefficients matrix (a) for space-by-time decomposition. b Muscle coherence networks for the multidimensional decomposition of EMG data in the spatiofrequency domain47. The inter-muscular coherence spectra of all muscle combination (m1-m2, m1-m3, …) are decomposed using non-negative matrix factorization. Each factor is characterized by the extracted feature (c) and the loadings of this feature in original spectra (w). These loadings give the strength of the edges between the nodes of each muscle network (right panel).
