Integrating No.3 lymph nodes and primary tumor radiomics to predict lymph node metastasis in T1-2 gastric cancer

BMC Medical Imaging

Table 2 Performance evaluation of models

Cohort	Models	TP	TN	FN	FP	Acc	Sen	Spe	PPV	NPV	AUC (95% CI)
Training cohort	R1	18	44	4	14	0.775	0.818	0.759	0.563	0.917	0.831 (0.725–0.937)
	R2	14	46	8	12	0.750	0.636	0.793	0.538	0.852	0.761 (0.629–0.893)
	CTR	10	54	12	4	0.800	0.455	0.931	0.714	0.818	0.693 (0.581–0.804)
	R1 + CTR	16	51	6	7	0.837	0.727	0.879	0.695	0.895	0.869 (0.789–0.949)
	R2 + CTR	16	46	6	12	0.775	0.727	0.793	0.571	0.885	0.814 (0.704–0.925)
	Nomogram	20	54	2	4	0.925	0.909	0.931	0.833	0.964	0.915 (0.832–0.998)
Testing cohort	R1	12	54	5	8	0.835	0.706	0.871	0.600	0.915	0.852 (0.742–0.962)
	R2	12	42	5	20	0.684	0.706	0.677	0.375	0.894	0.763 (0.626–0.900)
	CTR	6	58	11	4	0.810	0.353	0.935	0.600	0.841	0.644 (0.523–0.765)
	R1 + CTR	14	48	3	14	0.784	0.823	0.774	0.500	0.941	0.863 (0.772–0.954)
	R2 + CTR	14	45	3	17	0.747	0.824	0.726	0.452	0.938	0.753 (0.618–0.889)
	Nomogram	15	56	2	6	0.899	0.882	0.903	0.714	0.966	0.908 (0.814–1.000)

R1 radiomic signature-1, R2 radiomic signature-2, CTR CT-reported LN metastasis status, TP true positive, TN true negative, FN false negative, FP false positive, Acc accuracy, Sen sensitivity, Spe specificity, PPV positive predictive value, NPV negative predictive value, AUC area under curve, CI confidence interval

ISSN: 1471-2342