Early prediction of radiotherapy-induced parotid shrinkage and toxicity based on CT radiomics and fuzzy classification

Highlights：

• This paper aims at classifying patients, under radiotherapy for head and neck cancer, at risk of parotid gland shrinkage and 12 months xerostomia.

• Knowledge is extracted by means of Likelihood-Fuzzy Analysis, representing statistical information by fuzzy rule-based models.

• Clinical features, dosimetric parameters, and measures obtained by texture analysis of CT imagesare extracted to characterize real patients.

• For parotid shrinkage,we detectgood predictors and the models to use at different treatment stages, showing predictor-outcome dependencies.

• For 12-months toxicity, some possible predictors are detected, and the relation between final parotid shrinkage rate and 12-months xerostomia is evaluated.