Automated Detection of Myocardium Boundary in Rb-82 PET Images Using Wavelet Based Approach

Determination of ventricular boundary in myocardial perfusion images is essential for automated quantitation of perfusion and functional parameters. PET and PET/CT with higher spatial resolution produce more details with myocardial perfusion that can be utilized to improve the quantitative assessment with myocardial perfusion imaging. We propose a novel boundary detection algorithm which tracks the details of myocardium structure while maintaining the continuity and smoothness of epicardial and endocardial surfaces. This method was then tested on a cardiac phantom and clinical Rb-82 PET patients to demonstrate its potential for clinical utilization.