Table 7.
Outcomes of segmentation for various algorithms following the ablation study utilizing test datasets
| Algorithms | DSC(%) | IoU(%) | Precision(%) | Recall(%) |
|---|---|---|---|---|
| TransUNet | 77.31% | 77.80% | 76.4% | 73.12% |
| Swin-UNETR | 80.24% | 81.3% | 80.45% | 79.9% |
| MedT (Gated Axial Transformer) | 80.6% | 80.21% | 78.91% | 79.8% |
| DE-ResUNet | 82.3% | 82.6% | 81.59% | 80.5% |
| Hybrid CNN-ViT (ResNet-ViT) | 85.09% | 84.92% | 83.70% | 82.03% |
| ViT-Caps | 90.74% | 90.2% | 90.45% | 89.41% |
| Proposed Model | 98.5% | 97.5% | 97.07% | 98.65% |