Table 4.
Performance comparison between the proposed MGRL and several state-of-the-art methods for ASD vs. NC classification with fMRI data from ABIDE.
| Method | Algorithm | Accuracy | Recall | Precision | AUC | Sample # | Brain Atlas |
|---|---|---|---|---|---|---|---|
| MGRL (Ours) | GCN | 0.795 | 0.809 | 0.823 | 0.886 | 184 | AAL, CC200 |
The results of multi-scale graph representation learning (MGRL) are based on the NYU site. #represents the number of samples. The bold values mean to highlight the experiment results.