Figure 1.

Flow chart of the presented study. Image data from eight volunteers acquired with GE, Philips and Siemens scanners are used to assess the reproducibility of 411 radiomics features extracted on twelve imaging planes across the thoracic aorta throughout the cardiac cycle. The 51 most robust features are extracted for age- and sex- stratified cohorts of aortic stenosis patients and subjects without valve diseases for comparison. Age-matched cohorts are used to assess whether differences in flow profile are identifiable in the radiomics signatures. Finally, a classifier to differentiate healthy subjects from aortic stenosis patients is trained on the radiomics features of a cross-section in the ascending aorta (A1.1) for a peak systolic timeframe.