Automation Radiomics in Predicting Radiation Pneumonitis (RP)

Radiomics has shown great promise in predicting various diseases. Researchers have previously attempted to include radiomics their automated detection, diagnosis, and segmentation algorithms, taking these steps based on the promising outcomes of radiomics-based studies. As a result increased attention given this topic, numerous institutions developed own software. These packages, other hand, been utilized interchangeably without regard for fundamental differences. The primary purpose study was explore benefits predictive model performance radiation pneumonitis (RP), which is most frequent side effect chest radiotherapy, through work, we deep learning that intends increase RP prediction by combining more data points digging deeper into data. In order evaluate popular machine models, radiographic characteristics were used, recorded important them. high dimensionality radiomic datasets major issue. method proposed use problems synthetic minority oversampling technique, used create balanced dataset leveraging suitable hardware open-source present assessed efficacy including logistic regression (LR), support vector (SVM), random forest (RF), neural network (DNN), utilizing specific features. findings indicate four models displayed satisfactory forecasting pneumonitis. DNN demonstrated highest area under receiver operating curve (AUC-ROC) value, 0.87, suggesting its superior capacity among considered. AUC-ROC values forest, SVM, 0.85, 0.83, 0.81, respectively.