Papillary Muscle Derived Radiomic Features Facilitate LVH Detection and Improve LVH Etiology Classification

Detecting left ventricular hypertrophy(LVH) is common in a clinical scenario, while a precise classification of LVH etiology is quite challenging. In this study, we applied radiomics methods to LVH detection and LVH etiology classification. Innovatively, we extracted radiomic features not only from the myocardium(MYO) but also includes features from the papillary muscle(PM). After strict feature selection and analysis, we found the incorporation of PM features improves our model performance compared with models that simply use MYO features.