Yazdani Ladan, Bhatt Manish, Rafati Iman, Tang An, Cloutier Guy
IEEE Trans Ultrason Ferroelectr Freq Control. 2022 Jun;69(6):2061-2074. doi: 10.1109/TUFFC.2022.3166448. Epub 2022 May 26.
Ultrasound (US) shear wave (SW) elastography has been widely studied and implemented on clinical systems to assess the elasticity of living organs. Imaging of SW attenuation reflecting viscous properties of tissues has received less attention. A revisited frequency shift (R-FS) method is proposed to improve the robustness of SW attenuation imaging. Performances are compared with the FS method that we originally proposed and with the two-point frequency shift (2P-FS) and attenuation measuring US SW elastography (AMUSE) methods. In the proposed R-FS method, the shape parameter of the gamma distribution fitting SW spectra is assumed to vary with distance, in contrast to FS. Second, an adaptive random sample consensus (A-RANSAC) line fitting method is used to prevent outlier attenuation values in the presence of noise. Validation was made on ten simulated phantoms with two viscosities (0.5 and 2 Pa [Formula: see text]) and different noise levels (15 to -5 dB), two experimental homogeneous gel phantoms, and six in vivo liver acquisitions on awake ducks (including three normal and three fatty duck livers). According to the conducted experiments, R-FS revealed mean reductions in coefficients of variation (CV) of 62.6% on simulations, 62.5% with phantoms, and 62.3% in vivo compared with FS. Corresponding reductions compared with 2P-FS were 45.4%, 77.1%, and 62.0%, respectively. Reductions in normalized root-mean-square errors for simulations were 63.9% and 48.7% with respect to FS and 2P-FS, respectively.
超声（US）剪切波（SW）弹性成像已在临床系统中得到广泛研究和应用，用于评估活体器官的弹性。反映组织粘性特性的SW衰减成像受到的关注较少。本文提出了一种改进的频移（R-FS）方法，以提高SW衰减成像的稳健性。将其性能与我们最初提出的FS方法、两点频移（2P-FS）方法以及衰减测量超声SW弹性成像（AMUSE）方法进行了比较。在所提出的R-FS方法中，与FS方法不同，假设拟合SW频谱的伽马分布的形状参数随距离变化。其次，采用自适应随机样本一致性（A-RANSAC）直线拟合方法，以防止在存在噪声的情况下出现异常衰减值。在十个具有两种粘度（0.5和2 Pa[公式：见正文]）和不同噪声水平（15至-5 dB）的模拟体模、两个实验性均匀凝胶体模以及对清醒鸭进行的六次体内肝脏采集（包括三个正常鸭肝脏和三个脂肪肝肝脏）上进行了验证。根据所进行的实验，与FS方法相比，R-FS方法在模拟中显示变异系数（CV）平均降低了62.6%，在体模中降低了62.5%，在体内降低了62.3%。与2P-FS方法相比，相应的降低分别为45.4%、77.1%和62.0%。相对于FS和2P-FS方法，模拟中归一化均方根误差的降低分别为63.9%和48.7%。
