To investigate the potential of apparent diffusion coefficient (ADC) map-based deep learning and dose distribution-based dosiomics in predicting radiation-induced temporal lobe injury (RTLI) in nasopharyngeal carcinoma (NPC). This retrospective study included 3578 NPC patients from Jiangsu Cancer Hospital receiving intensity-modulated radiation therapy (IMRT). Ninety-four RTLI patients were recruited based on inclusion criteria and matched 1:1 with 97 control subjects using propensity scores. Patients were randomly assigned to the training cohort ( n = 135) and the validation cohort ( n = 59). Deep transfer learning (DTL) features and dosiomics features were extracted from ADC map and three-dimensional dose distribution, respectively. Pearson’s correlation coefficient and the least absolute shrinkage and selection operator (LASSO) regression were employed to identify predictive features. Subsequently, eight machine learning classification models were trained to establish a prediction framework, encompassing Support Vector Machine, K-Nearest Neighbor, Random Forest, Extremely Randomized Trees, eXtreme Gradient Boosting, Light Gradient Boosting Machine, Adaptive Boosting and Multilayer Perceptron. The performance of clinical, DTL, dosiomics and feature fusion model was compared by the area under the curve (AUC). We constructed six pre-trained transfer learning networks and extracted DTL features, respectively. The results showed that pre-trained WideResNet 101 exhibited superior performance with an AUC of 0.786 in the validation cohort. The clinical model based on D 1cc and induction chemotherapy demonstrated an AUC of 0.794 and the dosiomics model demonstrated an AUC of 0.903. Features fusion model demonstrated the highest AUC values in both the training (0.988) and validation (0.940) cohorts. The fusion model based on pretreatment ADC map and dose distribution provided a promising way to predict RTLI in NPC patients receiving IMRT, which can support clinicians in making decisions to develop individualized treatment plans and implement preventive measures.
Wang et al. (Wed,) studied this question.