Automatic 2D segmentation of the left ventricle in tagged cardiac MRI using motion information

A new algorithm is presented to automatically segment the left ventricle from tagged MR images using motion information. This segmentation method is based on a deformable template which is fitted to tags' envelope image through an iterative process. Motion information is used for initializing the template. The method has been evaluated both on simulated and 5 in vivo datasets. A 5 slice levels, 36 frames per slice dataset is fully segmented in less than 10 minutes. The method proposed here has the potential to overcome the main drawbacks of previous approaches, such as user supervision and computing time, to foresee routine clinical use of tagged cardiac MRI.