Fig. 2.

hMFA solution of Use Case 1. In Use Case 1, the hierarchical tree (a) was defined as in Fig. 1. From the global solution, we retained the first four components based on their percentage of variance accounted for (b). Comp. 1 explained 21.08%, Comp. 2 15.53%, Comp. 3 12.12%, and Comp. 4 10.22% of the total variance. The contribution of each layer into the definition of the components can be extracted, aiding with the interpretation of the results. For example, at the lumbar segment, the contribution of the variables conforming each spinal level show that the first component was most related to L3L4 and L4L5 variables, while L2L3 dominated in the second component (c). Equally, the contribution of the group of variables from the same muscle to the definition of the spinal level layer can be interpreted (d). In addition, the loadings (correlation between each variable and component) also aids interpretability by focusing on those loadings that has major absolute value (|loading|> 0.5) (e). MF: multifidus, Ps: psoas, ES: erector spinae, QL: quadratus lumborum