Skip to main content
. 2023 Aug 18;6:17. doi: 10.1186/s42492-023-00144-5

Fig. 4.

Fig. 4

Training and testing of one-fold of LOOCV for the ResNet18, VGG16 and ResNet50 models. In each LOOCV fold, the PET image slices of a patient were used as testing samples (right) and those of the other 80 patients were used as training samples (left)