Table 5. Classification results for two classes—early and late stages.

The best results for each test metric is indicated in bold italics.

Model	Train accuracy	Validation accuracy	Test accuracy	Test sensitivity	Test specificity
Siamese 3D	0.8542	0.6875	0.6473	0.6292	0.6292
ViT	0.7920	0.6990	0.6540	0.6560	0.6520
DenseNet	1.0000	0.7051	0.6540	0.6505	0.6505
EfficientNet B7	0.7760	0.7246	0.6652	0.6550	0.6550