Early classification of parotid glands shrinkage in radiotherapy patients: A comparative study

摘要: During radiotherapy treatment of patients with head-and-neck cancer, the possibility that parotid glands shrink was evidenced, connected increasing risk acute toxicity. In this ambit, early identification in danger is primary importance, order to treat them adaptive therapy. This work studies different approaches for classifying gland samples, taking into account textural features extracted from computed tomography (CT) images monitored patients. A real dataset used, and accuracy, sensitivity specificity are counted as classification performances. Therefore, firstly, procedures define classes compared terms their physical meaning Then, methods extracting knowledge implemented performances model interpretability. First-rate performance obtained by using Likelihood-Fuzzy Analysis (LFA), which a recently developing method based on use statistical information means Fuzzy Logic. The interpretable models LFA also allow identifying among those able predict shrinkage. Some these already known confirmed here, others new, some very predictors. Finally, an example feature monitoring patient presented, through reasoning scheme similar human reasoning, interpretation simple rule-based linguistic variables.