Radiomic Application In Predicting Short-term Efficacy Of Radiotherapy For Esophageal Carcinoma

Purpose:To forecast short-term efficacy of radiotherapy for esophageal carcinoma by extracting radiomic features from(computed-tomography)CT before treatoentMethods:59 patients diagnosed as esophageal carcinoma who received radiotherapy were enrolled in this study retrospectively.All patients received localization CT before radiotherapy and completed the whole course of radiotherapy.CT was rechecked within 1-3 months after radiotherapy.The shrinkage of tumors was used as a criterion for evaluating therapeutic effect.CT images were transmitted to the(treatment planning system)TPS,and the regbns of interest(ROI)were identified and delineated by two experienced radiotherapists.The outlined ROI is used for radoimics analysis,and the image features of ROI are extracted by using MATLAB software.The extracted features were analyzed statistically to find out the relationship between them and the shrinkage of tumors?Results:256 features were extracted fom each patient's CT.Spearman's rank correlation coe fficient showed that 28 image features were correlated with the efficacy(P<0.05),19 of which were positively correlated with the efficacy(r>0,P<0.05),and 9 were negatively correlated(r<0,P<0.05).Then,the short-term efficacy of radiotherapy was taken as dependent variable,and 32 indexes of sex,age,location,treatment plan and 28 image features were taken as independent variables.Finally,the linear regression relationship between the three image features and the short-term efficacy was obtained.Conclusions:Some image features are correlated with the degree of tumor regression after radiotherapy,which can be used to preliminarily forecast the short-term efficacy of esophageal carcinoma patients after radiotherapy.