CT-based delta-radiomics nomogram for predicting post-neoadjuvant chemoradiotherapy lymph node metastasis in esophageal squamous cell carcinoma: A feasibility study.
This study developed a nomogram using computed tomography (CT)-based delta-radiomics features and clinicopathological factors to predict lymph node metastasis (LNM) in patients with esophageal squamous cell carcinoma (ESCC) receiving neoadjuvant chemoradiotherapy (nCRT).
This study retrospectively enrolled 170 patients with ESCC receiving nCRT. The delta-radiomics signature model was constructed utilizing least absolute shrinkage and selection operator regression, and the radiomics score (radScore) was determined for each patient. A combined nomogram was established using the radScore and independent influencing factors obtained through univariate and multivariate analyses. The consistency and predictive ability of the nomogram were assessed using the calibration curve and the area under the receiver operating factor curve (AUC). The clinical benefits were assessed using decision curve analysis (DCA).
Two predictive models were constructed. The AUC values for the delta-radiomics signature model were 0.881 [95% confidence interval (CI): 0.827-0.935]. According to the univariate and multivariate analyses, the tumor length, tumor differentiation, and radScore were independent factors influencing LNM (P < 0.05). A combined nomogram was constructed from these factors, and the AUC reached 0.938 (95% CI: 0.898-0.979). DCA demonstrated that the clinical benefits of the nomogram for patients across an extensive range were more significant than the radiomics model alone.
This CT-based delta-radiomics nomogram model could benefit LNM in patients with ESCC following nCRT.
This study retrospectively enrolled 170 patients with ESCC receiving nCRT. The delta-radiomics signature model was constructed utilizing least absolute shrinkage and selection operator regression, and the radiomics score (radScore) was determined for each patient. A combined nomogram was established using the radScore and independent influencing factors obtained through univariate and multivariate analyses. The consistency and predictive ability of the nomogram were assessed using the calibration curve and the area under the receiver operating factor curve (AUC). The clinical benefits were assessed using decision curve analysis (DCA).
Two predictive models were constructed. The AUC values for the delta-radiomics signature model were 0.881 [95% confidence interval (CI): 0.827-0.935]. According to the univariate and multivariate analyses, the tumor length, tumor differentiation, and radScore were independent factors influencing LNM (P < 0.05). A combined nomogram was constructed from these factors, and the AUC reached 0.938 (95% CI: 0.898-0.979). DCA demonstrated that the clinical benefits of the nomogram for patients across an extensive range were more significant than the radiomics model alone.
This CT-based delta-radiomics nomogram model could benefit LNM in patients with ESCC following nCRT.