Table 4.
Evaluation results of the four best validation models on the MLD test set; the best metrics are highlighted in bold
| MLD | DSCM | DSCL | ASDM | ASDL | PREM (%) | PREL (%) | SENM (%) | SENL (%) | SPEM (%) | SPEL (%) |
|---|---|---|---|---|---|---|---|---|---|---|
| U-Net | 0.724 | 0.468 | 11.35 | 5.64 | 68.5 | 49.0 | 79.9 | 46.2 | 99.9906 | 99.9958 |
| V-Net | 0.766 | 0.482 | 2.38 | 4.34 | 73.6 | 39.2 | 83.1 | 67.2 | 99.9925 | 99.9902 |
| UNETR | 0.727 | 0.497 | 3.70 | 7.59 | 77.0 | 48.9 | 71.4 | 52.1 | 99.9946 | 99.9952 |
| SwinUNETR | 0.761 | 0.562 | 5.86 | 3.45 | 74.1 | 56.0 | 80.8 | 58.3 | 99.9928 | 99.9960 |
CSD Cerebrovascular segmentation dataset, MLD Middle cerebral artery M1 segment with lenticulostriate artery segmentation dataset, DSC Dice similarity coefficient, ASD Average surface distance, PRE Precision, SEN Sensitivity, SPE Specificity, *M Metrics of middle cerebral artery M1, *L Metrics of lenticulostriate artery