Evaluation of cardiac function from short-axis cardiac magnetic resonance images using automatic boundary extraction

Magnetic resonance images of the heart provide a wide variety of cardiac anatomical and physiological data. In this study an almost fully automatic process, that extracts the endocardial and epicardial boundaries and provides numerical values of the main cardiac functional parameters, is presented. The contour extraction algorithm performs well in more than 60% of the inspected images for variable quality images. Its performance improves markedly when only reasonable quality images are considered. The values obtained for cardiac parameters seem to be within the normal range