Automation analysis of cardiac wall deformation from tagged magnetic resonance images

Automation of cardiac wall deformation analysis from tagged magnetic resonance images (MRI) derives, basically, from the automatic detection of MR tags and left ventricle contours. In this work, it was adopted an approach based on image processing techniques and fuzzy logic to extract and classify image features as belonging to tags or ventricular borders. The use of fuzzy logic and IF-THEN rules, which involve image features such as length and curvature of valleys and gradients, allow the estimation of the membership of the pixels in the searched classes. The myocardial deformation is estimated in regions circumvented by contiguous tag intersections. The proposed method was applied to cine SPAMM (Spatial Modulation of Magnetization) short-axis images of the left ventricle obtained from human volunteers. (author)