Preoperative prediction of clinical and pathological stages for patients with esophageal cancer using PET/CT radiomics

Insights into Imaging

Table 2 The performance of the T stage, N stage, and pstage model for the training and validation cohorts

Stage	Model	Training cohort				Validation cohort
Stage	Model	AUC (95% CI)	Accuracy	Specificity	Sensitivity	AUC (95% CI)	Accuracy	Specificity	Sensitivity
T stage	CT_LR_Score	0.871 (0.792–0.951)	0.814	0.771	0.857	0.846 (0.702–0.991)	0.833	0.769	0.882
	PET_LR_Score	0.822 (0.725–0.920)	0.757	0.743	0.771	0.756 (0.576–0.936)	0.733	0.692	0.765
	Fused_LR_Score	0.675 (0.549–0.801)	0.643	0.600	0.686	0.665 (0.460–0.870)	0.700	0.538	0.824
	CT + PET_LR_Score	0.902 (0.834–0.971)	0.857	0.800	0.914	0.815 (0.646–0.983)	0.833	0.769	0.882
N stage	CT_LR_Score	0.883 (0.806–0.961)	0.829	0.821	0.839	0.769 (0.600–0.939)	0.733	0.647	0.846
	PET_LR_Score	0.780 (0.674–0.886)	0.700	0.564	0.871	0.760 (0.566–0.954)	0.867	0.824	0.769
	Fused_LR_Score	0.702 (0.579–0.825)	0.671	0.744	0.581	0.665 (0.456–0.874)	0.700	0.824	0.538
	CT + PET _LR_Score	0.918 (0.852–0.985)	0.960	0.923	0.871	0.824 (0.674–0.973)	0.733	0.647	0.846
pstage	CT_LR_Score	0.836 (0.745–0.927)	0.771	0.737	0.812	0.727 (0.542–0.912)	0.667	0.579	0.818
	PET_LR_Score	0.803 (0.701–0.904)	0.757	0.658	0.875	0.785 (0.611–0.958)	0.767	0.737	0.818
	Fused_LR_Score	0.706 (0.580–0.831)	0.729	0.842	0.594	0.689 (0.496–0.882)	0.700	0.526	1.000
	CT + PET _LR_Score	0.865 (0.780–0.950)	0.814	0.737	0.906	0.837 (0.689–0.986)	0.833	0.842	0.818