Table 1.
Dice similarity coefficient and Jaccard index when computed against MedSeg.
| MD 1 | MD 2 | |||||||
|---|---|---|---|---|---|---|---|---|
| Dice | % Diff * | Jaccard | % Diff * | Dice | % Diff * | Jaccard | % Diff * | |
| ResNet-SegNet | 0.77 | 1% | 0.63 | 2% | 0.74 | 4% | 0.60 | 5% |
| PSPNet | 0.79 | 4% | 0.65 | 5% | 0.77 | 0% | 0.64 | 2% |
| VGG-SegNet | 0.79 | 4% | 0.66 | 6% | 0.80 | 4% | 0.68 | 8% |
| VGG-UNet | 0.80 | 5% | 0.67 | 8% | 0.78 | 1% | 0.65 | 3% |
| ResNet-UNet | 0.83 | 9% | 0.71 | 15% | 0.80 | 4% | 0.68 | 8% |
| Mean of AI | 0.80 | 5% | 0.66 | 7% | 0.78 | 3% | 0.65 | 5% |
| MedSeg | 0.76 | - | 0.62 | - | 0.77 | - | 0.63 | - |
DSC: Dice similarity coefficient; JI: Jaccard index; * % Diff = absolute (COVLIAS − MedSeg)/MedSeg.