Robust segmentation of the left ventricle from MRI

Segmentation and extraction of cardiac structures from heart images has been of great importance in the quantification of indicative parameters of cardiac pathologies. In this work we propose an algorithm delineate cardiac cavities, based on a robust extraction of the left ventricle, together with a refined deformable model, that integrates edge confidence information. The final contours were compared with manual segmentations taking as a similarity measure the Tanimoto index; an average index of 0.74±0.10 was obtained. The estimated left ventricle volume profile along a cardiac cycle was also compared.