Automatic boundary detection of the left ventricle and measurement of local myocardial perfusion in MRI

Magnetic resonance imaging (MRI) is used to visualise myocardial function and perfusion. Functional images show the contraction of the heart while perfusion images show the blood flow in the myocardium. Extracting the left ventricle (LV) contours is an essential step for many quantitative analyses of the LV function and perfusion. In this study, both types of MR images are used and an automatic boundary detection and perfusion analysis algorithm is presented. Functional MR images present a better contrast between myocardium and cavities than perfusion MRI. Adding a functional slice to the perfusion series makes the automatic boundary detection easier. A ring representing the LV is automatically extracted and structured into six desired sectors, on which the perfusion analysis is concentrated. Our method is based on fuzzy clustering and fuzzy inference systems. The simplicity of the method makes the whole procedure real time