Table 2.
The features selected by the two-nest LOOCV loop.
| Name | The number of loops | P |
|---|---|---|
| gabor3_glszm_SZHGE | 117 | 0.007* |
| gabor7_glcm_cluster_shade | 123 | 0.247 |
| gabor14_glcm_cluster_tendency | 129 | 0.461 |
| gabor18_glcm_IMC2 | 122 | 0.591 |
| gabor18_glszm_LZLGE | 120 | 0.564 |
| gabor29_glszm_SZSE | 129 | 0.406 |
| gabor29_glszm_SZHGE | 125 | 0.375 |
| gabor30_glszm_SZHGE | 117 | 0.692 |
| gabor35_glszm_SZLGE | 129 | 0.510 |
| gabor36_glrlm_LGLRE | 119 | 0.979 |
| gabor36_glrlm45_LGLRE | 121 | 0.958 |
| gabor36_glszm_LGLZE | 129 | 0.402 |
| W5S5_fos_skewness | 123 | 0.486 |
| W5S5_fos_mass | 129 | 0.919 |
| W5W5_fos_mean | 128 | 0.431 |
| W5W5_fos_mass | 129 | 0.853 |
| R5S5_fos_median | 129 | 0.847 |
The number of loops indicates how many LOOCV loops have selected the feature. The P-value is calculated by the Student's T-test between the early- and late-onset of epilepsy patients.
*
Significant difference.