Computer Aided Effective Prediction of Complete Responders After Radiochemotherapy Based on Tumor Regression Grade Estimated by MR Imaging

The aim of this work is to implement an automatic method to predict and classify complete responders (CRs) patients, affected by rectal cancer and treated with neoadjuvant radiochemotherapy (RCT), by exploiting the tumor regression grade (MR-TRG) estimated by magnetic resonance imaging. For the purpose of the study, a total of 65 patients were enrolled and the magnetic resonance (MR) examinations to calculate TRG were performed using a 3.0 T scanner. By processing and testing patients’ data, the algorithm allows to determine the optimum threshold dividing CRs patients from patients that are considered non responders. The prediction accuracy of the classifier was investigated by using cross-validation statistical analysis in order to automatically determine the best testing rule. After collecting the outcomes of the performed cross-validation, the obtained results show the percentages of correct instances and misclassified patients. The automatic classification of CRs appears to be feasible and can be considered as a helpful method to predict CRs assisting clinicians to predict disease prognoses and patient survival prospects in order to provide treatments’ customization.