Table 1.
Demographic variables of the participants.
Table 2.
Feature Measures and Cortical Feature Index Information.
Fig 1.
Cortical segmentation based on the DKT atlas.
Each of the 31 anatomical ROIs are shown with a specific color and index (see Table 3 for indices). The figures on the left and right show the lateral and medial views, respectively.
Table 3.
DKT31 Protocol-based Cortical ROIs.
Fig 2.
Overall framework of the study.
The level-A block represents subject selection from the ADHD-200 dataset and briefly describes the preprocessing step, the level-B block provides information about the 31 DKT atlas-based ROIs, selection of the five feature measures and the total count of the cortical features. The level-C block elaborates the data validation methodologies and feature selection for the classifiers. Level-D represents the classifier choice for both the multiclass and binary settings.
Table 4.
Mean Multiclass (3 classes) classification results.
Table 5.
Binary classification results.
Fig 3.
Comparison of the testing accuracy of H-ELM, ELM, SVM linear, and SVM RBF in a multiclass setting using conventional validation and 10-fold cross-validation with RFE-SVM.
RCD represents the ranked classification dataset size acquired by cumulative RFE-SVM to achieve the highest accuracy.
Fig 4.
Twelve regions with significant differences as determined by ANOVA.
The left column shows the transverse view. The middle and right figures in the first row show the lateral view. The middle and right figures of the second row show the medial view. The most significant results are located in the superior frontal gyrus.
Table 6.
Most Significant Features for Classification.