Table 2.

Performance of radiomic models built by the MLR and SVM for the training and validation cohorts

Radiomic model			AUC (95% Cl)	Accuracy	Sensitivity	Specificity	PPV	NPV	F-1 score
T1WI model	The training cohort	MLR	0.85 (0.80–0.91)	0.81	0.82	0.80	0.83	0.78	0.82
		SVM	0.95 (0.92–0.99)	0.92	0.92	0.92	0.94	0.90	0.92
	The validation cohort	MLR	0.71 (0.81–0.91)	0.71	0.76	0.65	0.73	0.69	0.74
		SVM	0.85 (0.77–0.94)	0.74	0.71	0.76	0.79	0.68	0.75
T2WI model	The training cohort	MLR	0.87 (0.80–0.95)	0.83	0.88	0.77	0.83	0.83	0.85
		SVM	0.97 (0.95–0.99)	0.95	0.98	0.92	0.94	0.97	0.96
	The validation cohort	MLR	0.85 (0.90–0.94)	0.80	0.86	0.71	0.78	0.80	0.82
		SVM	0.74 (0.62–0.85)	0.68	0.76	0.59	0.70	0.67	0.73
T1-2WI model	The training cohort	MLR	0.95 (0.91–0.99)	0.86	0.90	0.82	0.86	0.86	0.88
		SVM	0.96 (0.92–0.99)	0.92	0.96	0.87	0.90	0.94	0.93
	The validation cohort	MLR	0.90 (0.85–0.95)	0.84	0.88	0.79	0.84	0.84	0.86
		SVM	0.93 (0.87–0.99)	0.87	0.81	0.94	0.94	0.80	0.87

PPV, positive predictive value; NPV, negative predictive value; T1WI, T1-weighted imaging; T2WI, T2-weighted imaging; SVM, Support vector machine; MLR, multivariable logistic regression