Developing Dosiomics Models for the Prediction of Postoperative Radiotherapy-Induced Esophagitis in Patients with Non-Small Cell Lung Cancer

Abstract Background: Radiotherapy-induced esophagitis (RE)diminishes quality of life and involves treatment interruption in patients with non-small cell lung cancer (NSCLC) undergoing postoperative radiotherapy. Dosimetric models showed limited capability in predicting RE. We aimed to develop dosiomics models to predict RE. Methods: Patients with NSCLC who underwent resection followed by postoperative radiotherapy between 2006 and 2015 were enrolled. The endpoint was grade ≥ 2 RE. Oesophageal three-dimensional dose distribution features were extracted using handcrafted and convolutional neural network (CNN)methods, screened using an entropy-based method, and selected using minimum redundancy maximum relevance. Prediction models were built using logistic regression. The area under the receiver operating characteristic curve (AUC) and precision-recall curve were used to evaluate prediction model performance. A dosimetric model was built for comparison. Results: Models were trained and validated using respective (n = 190) and prospective (n = 103) cohorts, respectively. Using handcrafted and CNN methods, 107 and 4,096 features were derived, respectively. Three handcrafted, four CNN-extracted, and three dosimetric features were selected. AUCs of training and test sets were 0.737 and 0.655 for the dosimetric features, 0.730 and 0.724 for handcrafted features, and 0.812 and 0.785 for CNN-extracted features, respectively. Precision-recall curves revealed that CNN-extracted features outperformed dosimetric and handcrafted features. Conclusions: Prediction models may identify patients at high risk of developing RE. Dosiomics models outperformed the dosimetric-feature model in predicting RE. CNN-extracted features were more predictive but less interpretable than handcrafted features.