Fig. 2From: Precision measurement of cardiac structure and function in cardiovascular magnetic resonance using machine learningStructure of the Unet used for short axis image segmentation. The model takes a grayscale CMR image with dimension 192 × 192 and creates a segmentation mask of the same dimension with 3 channels (one channel for each of: LV blood pool (white), myocardium (gray) and background (black)). The Unet used for long axis segmentations were the same, but image sizes and final layer were different—see Additional file 1: Table S1 for full detailsBack to article page