T2WI-based MRI radiomics for the prediction of preoperative extranodal extension and prognosis in resectable rectal cancer

Insights into Imaging

Table 3 Diagnostic performance of the models in the training cohort and validation cohort

Data set	Model	AUC (95% CI)	Sensitivity	Specificity
Training cohort	Intratumoral	0.618 (0.512–0.724)	0.659	0.571
	Intra-Peritumoral-MRF	0.638 (0.521–0.756)	0.793	0.486
	Intra-Peritumoral-3 mm	0.707 (0.612–0.803)	0.716	0.651
	Peritumoral 3 mm	0.626 (0.512–0.74)	0.435	0.812
	Peritumoral-MRF	0.614 (0.505–0.723)	0.475	0.730
	Clinical model	0.736 (0.641–0.832)	0.757	0.651
	Nomogram	0.799 (0.718–0.881)	0.676	0.837
Validation cohort	Intratumoral	0.612 (0.451–0.774)	0.382	0.688
	Intra-Peritumoral-MRF	0.586 (0.413–0.760)	0.676	0.625
	Intra-Peritumoral-3 mm	0.667 (0.493–0.841)	0.619	0.750
	Peritumoral 3 mm	0.625 (0.467–0.782)	0.645	0.421
	Peritumoral-MRF	0.550 (0.366–0.734)	0.75	0.286
	Clinical model	0.667 (0.434–0.899)	0.762	0.625
	Nomogram	0.723 (0.532–0.916)	0.548	0.875

MRF mesorectal fat, Intra-Peritumoral-MRF combined intratumoral with peritumoral-MRF, Intra-Peritumoral-3 mm combined intratumoral and peritumoral 3 mm