Merging single-track location Elastographic imaging with the frequency shift method improves shear wave attenuation measurements

Abstract

The frequency shift (FS) method is emerging as the standard approach for estimating shear wave attenuation coefficient (SWA). However, measurement noise can negatively impact the FS method’s accuracy, especially when employed in vivo. We hypothesized that combining plane wave single-track location shear wave elastography imaging with the FS method would reduce this problem. To test our hypothesis, we performed studies on calibrated phantoms and two groups of in vivo murine liver: control and obese mice. We evaluated the performance of various SWA methods, including the plane wave single-track location frequency shift (pSTL-FS) method that we recently developed, the original FS method, and the attenuation-measuring-shear-wave ultrasound elastography (AMUSE) method. We also assessed the effectiveness of assuming a Gaussian distribution versus a Gamma distribution for the shear wave spectrum when estimating SWA coefficients with the pSTL-FS and FS methods. The actual SWA coefficients of the phantoms were determined by performing independent mechanical testing on representative samples. The accuracy incurred when estimating SWA ranged from 84.69% to 97.55% for pSTL-FS (Gamma), 51.37%–72.18% for pSTL-FS (Gaussian), 40.33%–57.00% for FS (Gamma), 39.33%–55.37% for FS (Gaussian), and 59.25%–99.22% for AMUSE. The results of studies performed on murine livers (n = 10) revealed that assuming a Gaussian distribution during pSTL-FS imaging resulted in lower attenuation values than when a Gamma distribution was assumed. We also observed that pSTL-FS (Gamma) resulted in the highest significant difference between control and obese mice than all other approaches (p-value <0.0001). We also observed that the standard FS method with either Gamma or Gaussians produced lower attenuation estimates than pSTL-FS, AMUSE and mechanical testing. The mean attenuation coefficients of the murine livers measured with the pSTL-FS (Gamma and Gaussian functions) methods were consistently higher than those computed with the standard FS methods but lower than those computed with the AMUSE method. Our results demonstrated that combining the pSTL method with FS method provided more robust estimates of the SWA coefficient. For the murine livers, a Gamma distribution is more representative of the shear wave frequency spectrum than a Gaussian distribution.