Fig. 2
From: Effective workflow from multimodal MRI data to model-based prediction
Empirical and simulated features (connectome relationship) for machine learning. (a) Feature distributions across individual subjects for two brain parcellations given by the Schaefer atlas (100 regions) and the Harvard-Oxford atlas (96 regions) as indicated in the legend. (b,c) Principal component analysis (PCA) of the feature variability across 4 feature conditions: empirical, simulated and the two considered brain parcellations. The loadings and the fractions of the explained variance by different principal components are illustrated in plots (b) and (c), respectively. The color and line schemes are as in plot (a). The cumulative explained variance across all conditions is depicted in plot (c) by bars in light green color. eFC empirical functional connectivity, eSC empirical structural connectivity, sFC simulated FC, PC principal component.
