Table 5.
Validation of the proposed method on a secondary dataset (Souza et al., 2017)
| Dice | Sensitivity | Specificity | |
|---|---|---|---|
| ANTs | 95.93 (0.87) | 94.51 (1.58) | 99.71 (0.11) |
| BEaST | 95.77 (1.23) | 93.84 (2.57) | 99.76 (0.13) |
| BET | 95.22 (0.94) | 98.26 (1.61) | 99.13 (0.23) |
| BSE | 90.49 (7.03) | 91.44 (5.32) | 98.65 (2.27) |
| HWA | 91.66 (1.11) | 99.93 (0.12) | 97.83 (0.82) |
| MBWSS | 95.57 (1.46) | 92.78 (2.67) | 99.85 (0.04) |
| optiBET | 95.43 (0.71) | 96.13 (0.95) | 99.36 (0.31) |
| Proposed | 97.35 (0.44) | 97.72 (0.81) | 99.64 (0.16) |
| Proposed + EC | 97.58 (0.38) | 98.11 (0.80) | 99.65 (0.14) |
| ROBEX | 95.61 (0.72) | 98.42 (0.70) | 99.13 (0.28) |
| STAPLE | 96.80 (0.74) | 98.98 (0.60) | 99.38 (0.22) |
| Silver standard | 97.14 (0.51) | 96.83 (0.68) | 99.71 (0.11) |
The performances of other methods as reported in the original study are shown. The two top‐performing methods for each performance measure are emboldened. Note that both “Proposed” and “Proposed + EC” refer to the accelerated version of the method described in Section 3.1.