C. Pluempitiwiriyawej and S. Sotthivirat (Thailand)
Active contours, level sets, segmentation, myocardial per fusion, magnetic resonance imaging.
Quantitative analysis of myocardial perfusion re quires detection of myocardial boundaries through hun dreds of short-axis MR images. Manual tracing of my ocardial boundaries is a time-consuming and tedious task, which limits a clinical use of quantitative analysis. To fa cilitate for a more practical use, we propose an automatic algorithm for myocardial boundary detection in perfusion MR images. Our method is based on an active contour model called stochastic active contour scheme (STACS). However, we modify the original STACS to simultane ously perform two active contours for tracking both epi cardium and endocardium and segment an MR perfusion image into three different regions of interest: left ventri cle, myocardium and background. We assume each re gion has different statistics. Our results verify the sta tistical assumption of these three regions and show that our double-contour-STACS can successfully detect the my ocardial boundaries of a MR perfusion image.
Important Links:
Go Back
