Figure 2.

Brain imaging feature importance. The color bar indicates the feature importance that is the reduction of prediction accuracy (R ² of the quadratic regression model) with each feature removed from the DNN model. Brain features with positive contribution to age prediction include GMV widely distributed across the whole brain, FA and MD in the corpus callosum and cerebellar peduncle, ALFF in the frontal and occipital cortex, and ReHo in the frontal and temporal cortex. ReHo features in the parietal and occipital region contribute negatively to age prediction