We are training an AI to automatically detect and quantify the degree of incorrect rotation in left ventricular (LV) images within the apical four-chamber (A4C) view. To achieve this, we are ranking images based on the degree of rotation, from the most anticlockwise to the most clockwise. Quite often, there is very little difference in the degree of rotation between two images. When this happens, the exact order of ranking is not critical, just ensure that the images with significant differences in rotation are ranked correctly.