Improved Linear Contrast-Enhanced Ultrasound Imaging via Analysis of First-Order Speckle Statistics.

Abstract

The linear subtraction methods commonly used for preclinical contrast-enhanced imaging are susceptible to registration errors and motion artifacts that lead to reduced contrast-to-tissue ratios. To address this limitation, a new approach to linear contrast-enhanced ultrasound (CEUS) is proposed based on the analysis of the temporal dynamics of the speckle statistics during wash-in of a bolus injection of microbubbles. In the proposed method, the speckle signal is approximated as a mixture of temporally varying random processes, representing the microbubble signal, superimposed onto spatially heterogeneous tissue backscatter in multiple subvolumes within the region of interest. A wash-in curve is constructed by plotting the effective degrees of freedom (EDoFs) of the histogram of the speckle signal as a function of time. The proposed method is, therefore, named the EDoF method. The EDoF parameter is proportional to the shape parameter of the Nakagami distribution. Images acquired at 18 MHz from a murine mammary fat pad breast cancer xenograft model were processed using gold-standard nonlinear amplitude modulation, conventional linear subtraction, and the proposed statistical method. The EDoF method shows promise for improving the robustness of linear CEUS based on reduced frame-to-frame variability compared with the conventional linear subtraction time-intensity curves. Wash-in curve parameters estimated using the EDoF method also demonstrate higher correlation to nonlinear CEUS than the conventional linear method. The conceptual basis of the statistical method implies that EDoF wash-in curves may carry information about vascular complexity that could provide valuable new imaging biomarkers for cancer research.

• Authors: Matthew R Lowerison, M Nicole Hague, Ann F Chambers, James C Lacefield
• Journal: IEEE transactions on ultrasonics, ferroelectrics, and frequency control (IEEE Trans Ultrason Ferroelectr Freq Control) Vol. 63 Issue 9 Pg. 1409-1421 (09 2016) ISSN: 1525-8955 [Electronic] United States
• PMID: 27295664 (Publication Type: Journal Article, Research Support, Non-U.S. Gov't)