Table 1.
Evaluation of CAST automated segmentation in the MNI dataset.
Left/Right hippocampus | |||
---|---|---|---|
CAST DSC | 0.917(0.011)/0.919(0.013) | 0.890(0.017)/0.889(0.020) | 0.881(0.021)/0.875(0.025) |
Inter-rater DSCa | 0.903 (0.036) | 0.871 (0.053) | 0.876 (0.048) |
Intra-rater DSCa | 0.929 (0.010) | 0.905 (0.016) | 0.900 (0.019) |
CAST ICC | 0.912/0.914 | 0.887/0.886 | 0.880/0.875 |
Inter-rater ICCa | 0.91 | 0.73 | 0.90 |
Intra-rater ICCa | 0.91 | 0.94 | 0.96 |
CAST Jaccard coefficient | 0.848(0.019)/0.850(0.022) | 0.802(0.029)/0.800(0.032) | 0.787(0.032)/0.779(0.039) |
CAST | 0.906(0.011)/0.906(0.014) | ||
Generalized | |||
DSC |
The DSC and ICC scores of a single rater (intra-rater) and two independent raters (inter-rater) were directly extracted from Kulaga-Yoskovitz et al. (2015) and marked with gray background. All the rest metrics are calculated from CAST segmentation.