Fig. 1

Fully automated biventricular segmentation (upper panel) and manual segmentation (lower panel) of 6 representative cases. The figure comprises examples with good automated segmentation results (a-b) and limited automated segmentation results (c-f). Segmentation results of all phases and all short-axis slices can be found in the supplementary material. a End-diastolic phases of a patient with suspected myocarditis (1.5 T) and excellent biventricular segmentation. b End-systolic phases of a patient after transcatheter aortic valve replacement (TAVR) imaged at 3 T showing good segmentation results, except for insufficient segmentation of papillary muscles. c End-systolic phases of a patient with severe hypertrophic cardiomyopathy (HCM) imaged at 1.5 T with low accuracy of biventricular segmentations. d End-diastolic phases of a patient with dilated cardiomyopathy (DCM) imaged at 1.5 T with underestimation of right ventricular (RV) volume at the basal level. Also note misinterpretation of two apical thrombi as papillary muscles. e End-systolic phases of a patient with repaired Tetralogy of Fallot (ToF) with underestimation of RV volume due to severe metallic artefacts caused by sternal wires. f End-systolic phases of a patient with pulmonary arterial hypertension (PAH) with underestimation of RV volume most likely due to RV hypertrabeculation