Table 2 Performance of the 10 CPM-Net and 3D-CNN-TR based pipelines (with a fixed confidence threshold of 0.8) and post-processing methods 1–5, for the private and RSNA datasets.
None | Method 1 | Method 2 | Method 3 | Method 4 | Method 5 | |||
|---|---|---|---|---|---|---|---|---|
Private | CPM-Net | TP | 139 | 139 | 129 | 139 | 129 | 139 |
FP (FP/case) | 126 (0.88) | 98 (0.69) | 43 (0.30) | 48 (0.34) | 33 (0.23) | 37 (0.26) | ||
FN | 79 | 79 | 89 | 79 | 79 | 79 | ||
3D-CNN-TR | TP | 179 | 179 | 169 | 179 | 169 | 179 | |
FP (FP/case) | 182 (1.27) | 104 (0.73) | 98 (0.69) | 116 (0.81) | 79 (0.55) | 88 (0.62) | ||
FN | 39 | 39 | 49 | 39 | 39 | 39 | ||
RSNA | CPM-Net | TP | 791 | 790 | 691 | 767 | 690 | 766 |
FP (FP/case) | 748 (0.89) | 513 (0.61) | 499 (0.59) | 517 (0.61) | 304 (0.36) | 315 (0.37) | ||
FN | 236 | 237 | 366 | 260 | 337 | 261 | ||
3D-CNN-TR | TP | 940 | 936 | 829 | 917 | 827 | 914 | |
FP (FP/case) | 1,552 (1.84) | 1,028 (1.22) | 979 (1.16) | 1,178 (1.40) | 799 (0.95) | 872 (1.03) | ||
FN | 87 | 91 | 198 | 110 | 200 | 113 |
