Table 4.
Performance of the different automatic segmentation methods on the clinical routine dataset.
| Routine | DSC | Volume similarity | Volume error rate | Intraclass correlation | False positive rate | False negative rate |
|---|---|---|---|---|---|---|
| LPA |
0.652 (0.604–0.701) |
0.790 (0.733–0.846) |
1.011 (0.469–1.552) |
0.727 (0.546–0.836) |
0.402 (0.346–0.459) |
0.189 (0.149–0.229) |
| LGA | 0.490 (0.437–0.543) |
0.729 (0.664–0.794) |
2.533 (0.615–4.451) |
0.287 (0.050–0.497) |
0.560 (0.502–0.618) |
0.354 (0.305–0.403) |
| BIANCA | 0.607v(0.556–0.657) | 0.788 (0.733–0.843) |
0.709 (0.431–0.987) |
0.859 (0.774–0.913) |
0.404 (0.346–0.463) |
0.296 (0.247–0.344) |
| SLS | 0.613 (0.546–0.679) |
0.738 (0.676–0.801) |
0.515 (0.367–0.662) |
0.815 (0.708–0.885) |
0.289 (0.231–0.346) |
0.368 (0.288–0.448) |
| W2MHS | 0.223 (0.181–0.266) |
0.448 (0.382–0.515) |
0.682 (0.621–0.743) |
0.510 (0.157–0.719) |
0.461 (0.377–0.546) |
0.844 (0.812–0.877) |
| nicMSlesion(original) | 0.433 (0.377–0.489) |
0.647 (0.571–0.723) |
4.498 (0.667–8.33) |
0.396 (0.109–0.61) |
0.616 (0.558–0.674) |
0.351 (0.295–0.407) |
| nicMSlesion(retrained) | 0.500 (0.446–0.555) |
0.781 (0.722–0.841) |
1.349 (0.221–2.477) |
0.922 (0.868–0.954) |
0.505 (0.439–0.571) |
0.433 (0.39–0.476) |
| UBO | 0.560 (0.512–0.608) |
0.836 (0.789–0.882) |
0.569 (0.211–0.926) |
0.734 (0.584–0.834) |
0.471 (0.422–0.52) |
0.353 (0.301–0.405) |
For each metric, the table displays the average and the 95 % confidence interval within parentheses. DSC: Dice similarity coefficient. Results in bold indicates the best score for each metric.