Figure 3
Schematic diagram of the analysis pipeline. RsfMRI data from 4 sessions (R0, R1, R2, R3). Both rsfMRI from Visit 2 (real rTMS) and Visit 3 (sham rTMS) were used to compute RSFC-Snowballing density nodes and RSFC-Boundary Mapping gradients. Next, sham stimulation maps were subtracted from the corresponding real stimulation maps. The extracted ICA-based masks derived from the Visit 1 measurements were then applied to the corresponding node and gradient maps for feature selection. Finally, the remaining voxels were used for machine learning classification.
