Table 1.

AUC and specificity (with sensitivity = 0.95) of testing data for models with different combinations of features using RF with the complete training data.

Features	AUC	95% Confidence interval	Specificity	95% Confidence interval
GMM	0.88	0.840–0.913	0.49	0.373–0.609
ECC	0.82	0.775–0.865	0.30	0.203–0.420
GLCM	0.70	0.636–0.754	0.22	0.138–0.310
GMM + ECC	0.90	0.866–0.936	0.55	0.440–0.676
GMM + GLCM	0.89	0.851–0.921	0.55	0.425–0.640
ECC + GLCM	0.86	0.821–0.902	0.45	0.340–0.578
GMM + ECC + GLCM	0.91	0.873–0.941	0.62	0.514–0.736
CNN	0.93	0.904–0.957	0.69	0.600–0.787
ALL	0.94	0.913–0.961	0.72	0.612–0.803