A compound speckle model for vascular complexity quantification in nonlinear contrast-enhanced ultrasonography

A statistical method of contrast-enhanced ultrasound image analysis is introduced that models contrast speckle as a compound distribution of exponential probability density functions. The model, when applied to nonlinear destruction-reperfusion images acquired from a murine breast cancer xenograft model, generates a weighting function that is shown to distinguish between different vascular environments. Plotting the change in this weighting function over the course of microbubble reperfusion yields wash-in curves consistent with the expected mono-exponential kinematics. The proposed statistical method quantifies heterogeneities in contrast speckle from the tumor model and may provide useful new metrics for vascular characterization of tumors.