TABLE 1.
Mean and standard deviation of Dice, Jaccard index, and Hausdorff distance evaluating the RATS, SHERM, and 3D U-Net model for T2w images in different datasets.
| Dataset | Method | Dice | Jaccard index | Hausdoff distance |
| D1 | RATS | 0.9377 ± 0.0036*** | 0.8826 ± 0.0064*** | 8.97 ± 0.99*** |
| Sherm | 0.9767 ± 0.0028*** | 0.9545 ± 0.0054*** | 6.02 ± 1.32*** | |
| Proposed | 0.9898±0.0013 | 0.9800±0.0025 | 3.53±0.86 | |
| D2_PART1 (auditory) |
RATS | 0.9404 ± 0.0038*** | 0.8875 ± 0.0068*** | 8.77 ± 1.61*** |
| Sherm | 0.9686 ± 0.0079*** | 0.9392 ± 0.0147*** | 20.28 ± 8.55*** | |
| Proposed | 0.9842±0.0014 | 0.9690±0.0027 | 6.04±1.08 | |
| D2_PART2 (somatosensory) |
RATS | 0.9442 ± 0.0049*** | 0.8943 ± 0.0087*** | 9.69 ± 1.61*** |
| Sherm | 0.9735 ± 0.0048*** | 0.9484 ± 0.0091*** | 16.17 ± 6.04*** | |
| Proposed | 0.9845±0.0023 | 0.9696±0.0044 | 6.95±1.96 |
Bold values indicate the best results.
The asterisk (*) denotes a statistical significance when compared to the proposed method. ***p < 0.001.