Microbubble Signal Classification Using Normalized Singular Spectrum Area Based Filtering Methods

Targeted ultrasound contrast agents, comprising shell stabilized gas filled microbubbles (MBs), can be used to detect molecular markers of disease present on the vascular endothelium. Current MB imaging techniques exploit the nonlinear echo response of microbubbles to provide signal contrast with respect to adjacent tissue signal. However, these methods are hampered by false positive artifacts arising from nonlinear signals from strongly reflecting tissue interfaces. In this study, we demonstrate an image processing method that utilizes normalized singular spectrum area (NSSA) to distinguish adherent and non-adherent MB signals from static tissue signals with high specificity.