Application of radiomics in the differential diagnosis in ameloblastomas and dentigerous cysts. Part 2

Abstract

values of these lesions. We found statistically significant differences in 13 second-order features of dentigerous cysts and ameloblastomas, most of them were closely correlated. Multiple logistic regression analysis was performed to rank features and to determine most significant predictors. The final model included 2 features (Cluster shade and IMC 1) and provided high predictive value (the area under a ROC=0.93).