Table 4 Performance comparison of the survival prediction of convolutional neural network (CNN) models using positron emission tomography (PET) images
PET model | MAE (days) | C-index | Classification accuracy of 2-year survival status | Classification accuracy of 5-year survival status |
|---|---|---|---|---|
ResNet (MIP) | ||||
18 layers | 447 ± 26 | 0.710 ± 0.02 | 0.719 ± 0.03 | 0.920 ± 0.01 |
34 layers | 470 ± 14 | 0.713 ± 0.02 | 0.699 ± 0.03 | 0.908 ± 0.01 |
50 layers | 423 ± 22 | 0.717 ± 0.01 | 0.724 ± 0.03 | 0.924 ± 0.01 |
ResNet3D | ||||
10 layers | 440 ± 33 | 0.729 ± 0.01 | 0.726 ± 0.02 | 0.917 ± 0.01 |
18 layers | 429 ± 15 | 0.740 ± 0.01 | 0.733 ± 0.02 | 0.915 ± 0.01 |
34 layers | 405 ± 29 | 0.749 ± 0.02 | 0.751 ± 0.02 | 0.928 ± 0.01 |