Application Of CT Radiomics In Prediction Of Recurrence Of Locally Advanced Esophageal Squamous Cell Carcinoma After Trimodal Therapy

Part one:Application of CT Radiomics in Prediction of Early Recurrence of Locally Advanced Esophageal Squamous Cell Carcinoma after Trimodal TherapyObjective:This study aimed to develop models based on CT radiomics and clinical featuresof locally advancedesophageal squamous cell carcinoma(ESCC)to predict early recurrence.Methods:We collected electronic medical records and image data of 197 patients with confirmed locally advanced ESCC.These patients were randomly allocated to 137 patients in the training cohort and 60in the test cohort.352 radiomics features were extracted by delineating region-of-interest(ROI)around the lesion onCT images and clinical signature was generated by medical records.The radiomics model,clinical model,the combined model of radiomics and clinical features were developed by radiomics features and/or clinical characteristics.Predicting performance of the three models was assessed with area under receiver operating characteristic curve(AUC),accuracy and F-1 score.Results:Eleven radiomics features and/or six clinical signatures were selected to build prediction models related to early recurrence of locally advanced ESCC after trimodal therapy.The AUC of integration of radiomics and clinical models was better than that of radiomics or clinical model for the training cohort(0.821 versus 0.754 or 0.679,respectively)and for the validation cohort(0.809 versus 0.646 or 0.658,respectively).Integrated model of radiomics and clinical features showed good performance in predicting early recurrence of locally advanced ESCC for both the training and validation cohorts(accuracy=0.730 and 0.733,and F-1 score=0.730 and 0.7 78,respectively).Conclusion:The integrated model of CT radiomics and clinical features may be a potential imaging biomarker to predict early recurrence of locally advanced ESCC after trimodal therapy.Part two:CT Radiomics FeaturestoPredict the Recurrence ofLocally Advanced Esophageal Squamous Cell Carcinoma within Two Years after Trimodal TherapyObjective:To investigate whether radiomics features based onCT can predict recurrence of locally advanced ESCC within two years after trimodal therapy.Methods:This study retrospectively collected 220 patients with pathology-confirmed locally advanced ESCC(154 in the training cohort and 66 in the validation cohort).All patients underwent thoracoabdominal CT before trimodal therapy.352 radiomics features were extracted from CT data and clinical features were acquired from the clinical medical record system.Univariate statistical tests and the least absolute shrinkage and selection operator method(LASSO)were performed to select the optimal radiomics features.Logistic regression was conducted to build the radiomics model,clinical model,and the combined model ofboth radiomics and clinical features.The predictive performance was judged by the main predictors including AUC,accuracy and Fl-scores in the training and the validation cohorts.Results:Ten optimal radiomics features and/or seven clinical features were selected to build radiomics model,clinical model and the combined model.To predict recurrence within two years in locally advanced ESCC after trimodal therapy,the CT radiomics model yielded AUC values of 0.815 and 0.720,accuracy values of 0.773 and 0.700,and F1-scores of 0.839 and 0.800 in training and validation cohort,respectively.The clinical model yielded AUC values of 0.763 and 0.750,accuracy values of 0.740 and 0.697,F1-scores of 0.815 and 0.787 in the training and validation cohort,respectively.In comparisons with the previous two models,the combined model demonstrated best performance with AUC values of 0.879 and 0.857,accuracy values of 0.844 and0.788,and F1-scores of 0.886 and 0.851 in training and validation cohort,respectively.Conclusion:The radiomics featurescould have great potential to predict the recurrence of locally advanced ESCC within two years after trimodal therapy.And the combine model of radiomics and clinical features shows better performance than the radiomics model or clinical model.