A two-component speckle model for detection of microbubble signals in linear contrast-enhanced ultrasonography

Contrast-enhanced ultrasound (CEUS) serves oncology by imaging tumor blood supply to enable quantification of longitudinal vascular changes and monitoring of treatment responses. Unfortunately, the linear subtraction methods commonly used for preclinical imaging are susceptible to registration errors and motion artifacts that lead to reduced contrast-to-tissue ratios. In this presentation, an alternative approach is proposed to improve discrimination between the contrast and tissue signals by comparing the first-order speckle statistics of images acquired before and after injection of microbubbles. The microbubble signal component is modeled as a temporally varying random process superimposed on a Rayleigh-distributed speckle signal representing backscatter from tissue. Images were acquired at 18 MHz from a murine orthotopic (mammary fat pad) xenograft breast cancer model following a bolus injection of microbubbles. Images were processed using gold-standard pulse inversion nonlinear CEUS, conventional linear subtraction, and the proposed statistical method. In comparison to conventional linear CEUS, the statistical method produced a wash-in curve that showed closer agreement to the gold-standard nonlinear CEUS data. The statistical method eliminates baseline image subtraction from linear CEUS processing, which should streamline the imaging workflow, improve the robustness of image quantification, and enable real-time perfusion imaging with linear CEUS.