Multiparametric MRI-based intratumoral and peritumoral radiomics for predicting the pathological differentiation of hepatocellular carcinoma

Insights into Imaging

Table 4 Diagnostic performance of the different model for predicting pHCC in the training and validation cohorts

Model name	Dilation distance (mm)	Number of features	Cohort	Accuracy	AUC (95% CI)	Sensitivity	Specificity
Clinical	NA	3	Training	0.63	0.77 (0.69–0.86)	0.83	0.57
Clinical	NA	3	Validation	0.77	0.73 (0.45–1.00)	1.0	0.80
Intratumoral	NA	12	Training	0.85	0.92 (0.87–0.96)	0.83	0.85
Intratumoral	NA	12	Validation	0.85	0.82 (0.58–1.00)	1.00	0.90
Peritumoral_5mm	5	15	Training	0.85	0.84 (0.74–0.93)	0.77	0.87
Peritumoral_5mm	5	15	Validation	0.77	0.77 (0.52–1.00)	1.00	0.80
Peritumoral_10mm	10	7	Training	0.89	0.87 (0.80–0.95)	0.71	0.94
Peritumoral_10mm	10	7	Validation	0.62	0.80 (0.37–1.00)	1.0	0.55
Peritumoral_20mm	20	7	Training	0.76	0.73 (0.62–0.84)	0.66	0.79
Peritumoral_20mm	20	7	Validation	0.62	0.68 (0.32–1.00)	1.00	0.60
IntraPeri	NA	12	Training	0.81	0.95 (0.91–0.98)	0.97	0.77
IntraPeri	NA	12	Validation	0.77	0.86 (0.56–1.00)	1.00	0.73