Maria J. Ledesma-Carbayo1, Ana Bajo1, John Andrew Derbyshire2, Andres Santos1, Manuel Desco3, Elliot R. McVeigh4
1Ingeniera Electrnica, Universidad Politcnica de Madrid, Madrid, Spain; 2Laboratory of Cardiac Energetics, National Heart, Lung & Blood Institute, National Institutes of Health, Bethesda, MD, USA; 3Unidad de Medicina y Ciruga Experimental, Hospital General Universitario Gregorio Maraon, Madrid, Spain; 4Department of Biomedical Engineering, Johns Hopkins University School of Medicine, Baltimore, MD, USA
In this work, non-rigid image registration is used to estimate myocardial motion 3D Tagged MR datasets. The method proposed is a multi-source method that takes advantage of all the image information present in the short and long axis tagged images. Myocardial or tag segmentation is not required and therefore the method is completely unsupervised. The method has been evaluated with respect to the framework provided by the programs FindTags (tag segmentation) and Tag Tissue Tracker (motion field fitting). The evaluation was performed on experimental animal datasets achieving subpixel agreement between the fully automated tracking and the Findtags approach.