Figure 1

A) The average Dice score across all registrations, all frames (and both manual segmentations) is shown in the top left. A range of image metrics and registration algorithms were employed to identify an optimal combination. B) The Dice coefficient over the three heart beats is shown prior to (blue) and after registration (red). This result is obtained using the best registration identified across all 9 patients. C) The success of the registration is illustrated with the underlying images and transforms masks. The frame shown corresponds to the worst Dice score prior to registration. As you can see, both the transformated image and mask show close agreement. D) We are interested in measuring wall thickness. Here we illustrate how this algorithm allows us to observe the underlying three heartbeats by showing a cross section of the myocardial mask.