Table 2.

Comparison of classification accuracy between CNNs in the EBV + MSI/others task. Each value shows the average accuracy (the average ratio of correct prediction for each 224 × 224 pixels image, ranging from 0.0 to 1.0), obtained by fourfold cross-validation. The highest accuracy is indicated in bold. CNN, Convolutional neural networks; EBV, Epstein-Barr virus; MSI, Microsatellite instability.

CNN architecture	Validation accuracy
VGG16-based	0.828
VGG19-based	0.818
ResNet50-based	0.811
EfficientNetB0-based	0.827