Computed tomography-based radiomics nomogram for prediction of lympho-vascular and perineural invasion in esophageal squamous cell cancer patients: a retrospective cohort study

Table 2 Diagnostic performance of LVI and PNI prediction models in training and validation groups

Prediction Model/Evaluation Metric		TPR	FPR	TNR	FNR	Accuracy	F1-Score	AUC(95%CI)
LVI Radiomics Signature	Training group	0.89	0.46	0.54	0.11	0.61	0.47	0.77 (0.71 - 0.82)
LVI Radiomics Signature	Validation group	0.60	0.19	0.81	0.40	0.77	0.56	0.74 (0.64 - 0.84)
LVI Nomogram	Training group	0.76	0.18	0.82	0.24	0.81	0.63	0.82 (0.77 - 0.87)
LVI Nomogram	Validation group	0.74	0.33	0.77	0.26	0.76	0.53	0.80 (0.71 - 0.88)
PNI Radiomics Signature	Training group	0.69	0.39	0.61	0.31	0.63	0.49	0.69 (0.63 - 0.75)
PNI Radiomics Signature	Validation group	0.71	0.38	0.62	0.29	0.64	0.43	0.68 (0.58 - 0.77)
PNI Nomogram	Training group	0.73	0.33	0.67	0.27	0.68	0.51	0.75 (0.69 - 0.80)
PNI Nomogram	Validation group	0.74	0.34	0.66	0.26	0.68	0.52	0.72 (0.62 - 0.82)

TPR True Positive Rate, FPR False Positive Rate, TNR True Negative Rate, FNR False Negative Rate, LVI Lympho-vascular Invasion, PNI Perineural Invasion, AUC Area Under the Curve

ISSN: 1470-7330