Figure 1

Reading from left to right, the steps used by the segmentation algorithm to separate scar from viable myocardium. The example shows the optimal segmentation of a dual IR DE-CMR image with a T1,min of 200 ms. Note that the myocardial cross-sections that form the input to this program have already been manually extracted from the original DE-CMR image. Under each step representative myocardial cross-sections and image histograms are shown. Features coloured in: pink, relate to the viable myocardium; green, to the pixels with intensity greater than the threshold; orange, to the pixels with intensity greater than the threshold and which belong to regions identified as scar by the feature analysis step. The blue rings indicate the weighted centroid for each aggregate of bright pixels that remains after standard deviation thresholding. These positions are used for subendocardial distance feature analysis.