Skip to main content
. 2021 May 7;15:634063. doi: 10.3389/fnins.2021.634063

FIGURE 5.

FIGURE 5

Feature ranks and classification accuracy in lesion-NAWM analysis. Top panel: The mean (standard deviation) rankings of the 11 diffusion features based on the linear recursive feature elimination (RFE) algorithm in SVM. Bottom panel: Accuracy of the classification models (SVM-IC) obtained by adding features one at a time, starting from the highest to the lowest rankings. The stars indicate features that contribute to significant improvement in classification accuracy (****p < 0.0001).