Table 4.

Classification metrics of all tested models. Four radiomic models using different input features were compared with conventional measures. The respective performance metrics (Matthews correlation coefficient (MCC), balanced accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV)) determined on the test set are reported

Model	MCC	Balanced accuracy	Sensitivity	Specificity	PPV	NPV
LN short diameter	0.32	0.67	0.35	1.00	1.00	0.29
LN Volume	0.37	0.72	0.67	0.77	0.92	0.38
Expert rating	0.32	0.67	0.35	1.00	1.00	0.29
Radiomics-texture	0.30	0.68	0.73	0.62	0.88	0.38
Radiomics-shape	0.37	0.73	0.61	0.85	0.94	0.37
Radiomics-intensity	0.25	0.57	0.98	0.15	0.81	0.67
Radiomics-LBP	0.74	0.84	0.98	0.69	0.92	0.9
Radiomics-combined	0.64	0.73	1.00	0.46	0.88	1.00