Cardiac Magnetic Resonance Cine Images derived-Radiomics for the Prediction of Event Free Survival in Patients with Acute Myocardial Infarction

Motivation: Prognostic value of radiomic features extracted from CMR cine image remains to be investigated. Goal(s): To evaluate the prognostic value of radiomic features derived from cine images in patients with ST-segment elevation myocardial infarction (STEMI). Approach: Radiomic features were extracted from CMR cine images on STEMI patients, and LASSO -Cox regression used to select predictive features for MACE. Cox regression was applied to build models. Results: RAD score provided an incremental prognostic value above baseline clinical factors and LVEF (C-index 0.78 vs 0.69; p=0.002) and outperformed the addition of CMR markers of infarct injury (C-index: 0.78 vs 0.69, p<0.001). Impact: Radiomic features provide incremental prognostic value to clinical and infarct size in the prediction of MACE, which would promote the development of the prognostic assessment with non-contrast enhanced CMR.