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

Insights into Imaging

Table 4 Comparison of the diagnostic performance of the models in the training cohort and validation cohort

Data set	Models comparison	AUC	p
Training cohort	Intratumoral vs Intra-Peritumoral-MRF	0.618 vs 0.638	0.8
	Intratumoral vs Intra-Peritumoral-3 mm	0.618 vs 0.707	0.2
	Intratumoral vs clinical model	0.618 vs 0.736	0.1
	Intratumoral vs nomogram	0.618 vs 0.799	0.008
	Intra-Peritumoral-MRF vs Intra-Peritumoral-3 mm	0.638 vs 0.707	0.4
	Intra-Peritumoral-MRF vs clinical model	0.638 vs 0.736	0.2
	Intra-Peritumoral-MRF vs nomogram	0.638 vs 0.799	0.03
	Intra-Peritumoral-3 mm vs clinical model	0.707 vs 0.736	0.7
	Intra-Peritumoral-3 mm vs nomogram	0.707 vs 0.799	0.03
	Clinical model vs nomogram	0.736 vs 0.799	0.072
Validation cohort	Intratumoral vs Intra-Peritumoral-MRF	0.612 vs 0.586	0.8
	Intratumoral vs Intra-Peritumoral-3 mm	0.612 vs 0.667	0.3
	Intratumoral vs clinical model	0.612 vs 0.667	0.7
	Intratumoral vs nomogram	0.612 vs 0.723	0.4
	Intra-Peritumoral-MRF vs Intra-Peritumoral-3 mm	0.586 vs 0.667	0.2
	Intra-Peritumoral-MRF vs clinical model	0.586 vs 0.667	0.7
	Intra-Peritumoral-MRF vs nomogram	0.586 vs 0.723	0.4
	Intra-Peritumoral-3 mm vs clinical model	0.667 vs 0.667	0.8
	Intra-Peritumoral-3 mm vs nomogram	0.667 vs 0.723	0.9
	Clinical model vs nomogram	0.667 vs 0.723	0.4

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