204P A radiomics strategy based on CT intra-tumoral and peritumoral regions for preoperative prediction of neoadjuvant chemoradiotherapy for esophageal cancer

Although neoadjuvant chemoradiotherapy followed by surgery is the standard treatment for esophageal cancer patients, most patients are unable to achieve pathological complete response with neoadjuvant therapy, resulting in poor outcomes. The aim of this study is to develop a method for selecting patients who can achieve pathological complete response through pre-neoadjuvant therapy chest-enhanced CT scans. Two hundreds and one patients with esophageal cancer were enrolled and divided into a training set and a testing set in a 7:3 ratio. Radiomics features of intra-tumoral and peritumoral images were extracted from preoperative chest-enhanced CT scans of these patients. The features were dimensionally reduced in two steps. The selected intra-tumoral and peritumoral features, including marginal (with a distance of 0-3mm from the tumor) and adjacent (with a distance of 3-6mm from the tumor) ROI, were used to build models with four machine learning classifiers, including Support Vector Machine, XG-Boost, Random Forest and Naive Bayes. Models with satisfied accuracy and stability levels were considered to perform well. Finally, the performance of these well-performing models on the testing set was displayed using ROC curves. Among the 16 models, the best-performing models were the integrated (intra-tumoral and peritumoral features)-XGBoost and integrated-random forest models, which had average ROC AUCs of 0.906 and 0.918, respectively, with relative standard deviations (RSDs) of 6.26 and 6.89 in the training set. In the testing set, the AUCs were 0.845 and 0.871, respectively. There was no significant difference in the ROC curves between the two models.Table: 204PThe performance of the selected models on the testing setModelAUC (95% CI)SpecificitySensitivityIntegrated-XGBoost0.845 (0.764, 0.928)0.8640.777Original-XGBoost0.759 (0.660, 0.857)0.9000.592Integrated-Random Forest0.871 (0.796, 0.946)0.6820.933Original-Random Forest0.795 (0.703, 0.887)0.8250.673Adjacent-Random Forest0.769 (0.671, 0.868)0.8860.533Integrated-Support Vector Machine0.719 (0.613, 0.825)0.7950.622 Open table in a new tab The addition of peritumoral radiomics features to the radiomics analysis may improve the predictive performance of pathological response for esophageal cancer patients to neoadjuvant chemoradiotherapy.