Automatic segmentation performance on our testing data set (trained using distributionally robust optimization) compared with other methods trained and evaluated with different data setsa
| Our Data Set | Habas et al26 | Serag et al27 | Khalili et al11 | Gholipour et al4 | |
|---|---|---|---|---|---|
| Fetal data set | Abnormal | Normal | Normal | Not specified | Both normal and abnormal |
| WM | 96.2 (9.0) | 90.0 (2.0) | 90.0 (6.0) | 91.9 | |
| V | 93.7 (7.8) | 90.0 (2.0) | 92.0 (4.0) | 87.4 | 88.0 (6.7) |
| C | 94.7 (4.4) | 79.4 |
Note:—V indicates ventricular system; C, cerebellum.
↵a We report the mean DC (%). SD is noted in parentheses (when available). Performance of previous methods is taken from the literature. Hence, this comparison can be used as an indication only. Whether the test data set contains normal and/or abnormal fetal brains is specified in the row “Fetal data set.”