Table 4.
Top 21 image features that were used in the best support vector machine model.
| Feature type | Scan | Feature | Ranking |
|---|---|---|---|
| Pre-contrast | Skewness | 10 | |
| Histogram | Post-contrast 1 | Skewness | 2 |
| Post-contrast 2 | Skewness | 1 | |
| Post-contrast 3 | Skewness | 3 | |
| Post-contrast 1 | Kurtosis | 5 | |
| Post-contrast 3 | Kurtosis | 13 | |
| Pre-contrast | Minimum intensity | 8 | |
| Post-contrast 1 | Median intensity | 17 | |
| Shape | Pre-contrast | Solidity | 4 |
| Pre-contrast | Extent | 6 | |
| Pre-contrast | Eccentricity | 7 | |
| GLCM | Post-contrast 1 | Contrast | 9 |
| GLSZM | Pre-contrast | Large-zone emphasis (LZE) | 18 |
| Post-contrast 2 | Large-zone emphasis (LZE) | 14 | |
| Post-contrast 3 | Large-zone emphasis (LZE) | 15 | |
| Post-contrast 1 | Low-intensity zone emphasis (LIZE) | 21 | |
| Post-contrast 2 | Low-intensity zone emphasis (LIZE) | 19 | |
| Post-contrast 1 | High-intensity zone emphasis (HIZE) | 11 | |
| Post-contrast 2 | Low-intensity large-zone emphasis (LILZE) | 12 | |
| Post-contrast 3 | Low-intensity large-zone emphasis (LILZE) | 16 | |
| Post-contrast 2 | High-intensity large-zone emphasis (HILZE) | 20 |