2648

View Abstract / View Presentation

Deep-Learning epicardial fat quantification using 4-chambers Cardiac MRI segmentation, comparison with total epicardial fat volume

Pierre Daudé¹, Patricia Ancel², Sylviane Confort-gouny¹, Anne Dutour², Bénédicte Gaborit², and Stanislas Rapacchi¹
¹Aix-Marseille Univ, CNRS, CRMBM, Marseille, France, ²APHM, Hôpital Universitaire Timone, Service d’Endocrinologie, Marseille, France

Deep-learning segmentation of epicardial adipose tissue surface in 4CH cine proved the evaluation of this long-overseen biomarker feasible in a database of 126. Networks reached relative surface errors <20% within the upper half of the test set, when 2 observers agreed within 15%.

Figure 1 : Comparison of total epicardial fat volume against 4-chamber surface measured on systolic or diastolic frame across the three cohorts merged for the database.

Figure 3 : Representative automated segmentation results for each of EAT surface population quartile. White arrows shows network segmentation errors.