Table 2.

Detailed performance of all tested models. Siamese_cor, Siamese_sag, and Siamese_axial represent the models trained with coronal, sagittal and axial views respectively, and Siamese_majority is the model representing majority voting. SVM-radiomic represents the SVM model utilizing only radiomic features, and the ensemble model combines the SVM-radiomics and Siamese models.

Models		ACC	SEN	SPE	AUC
Siamese-network models	Siamese_cor	0.83	0.94	0.83	0.89
	Siamese_sag	0.86	0.97	0.85	0.91
	Siamese_axial	0.87	0.97	0.86	0.92
	Siamese_majority	0.88	0.97	0.88	0.93
SVM-Radiomics model		0.92	0.80	0.93	0.92
Ensemble model		0.91	0.96	0.90	0.93