山口 匡

Choosing an appropriate dynamic range for acquiring radio-frequency (RF) data from a high-frequency ultrasound (HFU) system is challenging because the RF data amplitude typically covers several orders of magnitude between the sample surface and the deepest imaged regions. In addition, the saturated signal may decrease the accuracy of QUS methods because quantitative ultrasound (QUS) methods are sensitive to saturated data. In this study, the effects of saturation on QUS estimates of Nakagami shape parameter (Nakagami parameter) were quantified by analyzing data acquired from 20 dissected human lymph nodes with a single-element transducer operating at a center frequency of 26 MHz. Artificially saturated signals (x(sat)) were produced by applying artificial saturation methods to the original unsaturated signals (x(ori)). Saturation degree was quantified using an index termed Saturate-SNR (SSNR). Nakagami parameters were estimated from xsat over a wide range of SSNR values. Nakagami parameters of saturated signals were increased (0.18 when the signal decremented 2 bit) significantly with decreasing SSNR. Nakagami parameters were corrected by pretreatment that applied a smoothing spline to the saturated signal. The smoothing spline restoration method is tuned P parameter. The best correction occurred when P was 1 (i.e., cubic spline interpolation). The maximum Nakagami parameter error in the corrected with 6-bit signal was 0.10, which is less than the average difference of 0.12 that existed between non-metastatic and metastatic lymph nodes.

We aim to characterize tissues of skin ulcers for bacterial infection quantitatively. The echo amplitude envelope of the backscattered signals can be described by some statistics distributions in consistent with the different scattering. The basic statistics distributions (Rayleigh, K, and Rician distributions) were used to detect the difference of backscattered signals between non-infected and infected tissue. Three kinds of animal skin ulcer models (non-infection, colonization, and infection models) with open injury in rats were characterized in this study. Ultrasound data were acquired by modified ultrasonic diagnostic equipment with a linear phased array transducer. The center frequency of a linear phased array ultrasound transducer was 8.9 MHz. In addition to animal models study, we simulated the echo signal from random scatterer mediums to confirm the relationship between variation of quantitative ultrasound (QUS) parameters and scatterer structure. As QUS parameters, the ratio of the mean to standard deviation of the echo amplitude envelope and the statistics distribution parameter were computed in each analysis. In the infection model rat, the typical difference of statistics distribution parameters (non-Rayleigh or Rayleigh distribution parameter) corresponding to the observed histopathological difference are shown in noninfected and infected tissues. Additionally, this tendency are described in 2D analysis of each model rat. From the result of computer simulation, scatterer number density is correlated with QUS parameters, and assumes to be also influenced by the histopathological structure in animal models whether infected or not.

A quantitative diagnostic method for liver fibrosis using ultrasound echo signals is highly required. A probability density function (PDF) of echo envelope from a normal liver can be approximated by a Rayleigh distribution; however, the PDF of echo envelope from liver fibrosis deviates from the Rayleigh distribution. To evaluate tissue characteristics in the ultrasound B-mode image, several amplitude distribution models have been proposed. We proposed a multi-Rayleigh distribution model and evaluation method of liver fibrosis using the multi-Rayleigh model. In this study, we evaluated the modeling accuracy of the multi-Rayleigh model and other amplitude distribution models using the KL divergence. From the evaluated results for the 120 clinical data, it was found that the multi-Rayleigh model with three components is more suitable model than other amplitude distribution models for approximating the PDF of echo envelope from the liver fibrosis.

皮膚下の感染に対する簡易かつ定量的な診断手法の1つとして、超音波定量診断法に基づく組織性状診断に着目した。本報告では、3種のラット皮膚潰瘍モデルを計測対象とし、病理学的評価と超音波顕微鏡を用いた音響特性評価を同時に行い、潰瘍内の顕微レベルでの特徴を理解した上で、皮膚下の組織構造をマクロ的に捉えた解析を行う。まず、超音波診断装置から取得したエコー信号の振幅包絡確率密度分布と複数の統計モデルの一致度を評価し、皮膚潰瘍内の散乱体構造を反映できる統計モデルについて検討した。さらに、感染により生体組織内の散乱体密度に変化が起こると想定し、その変化を反映できる統計モデルパラメータに着目した。その結果、非感染と感染における組織性状の差異に起因すると考えられる各統計モデルにおけるモデルパラメータの差異を確認した。(著者抄録)

Early diagnosis of non-alcoholic steatohepatitis (NASH) is highly desired because NASH can lead to cirrhosis or even to hepatocellular carcinoma in some severe cases. Towards non-invasive diagnosis with ultrasound, we studied free fatty acids (FFAs) present in the liver, which is the organ most likely to be affected by the disease. As a preclinical study, we performed acoustic-impedance measurements of five kinds of FFAs in solvent or in cultured Huh7 cells. To measure the acoustic impedance, a concave transducer with an 80-MHz center frequency was incorporated in a scanning acoustic microscopy system. One-way ANOVA showed statistically-significant differences (p<0.05) in acoustic impedance among the FFAs in FFA solvent and with cultured Huh7 cells. These results suggest that each of the FFAs, especially PA, OA and PAOA could be distinguished from each other regardless of whether they were in solution or in absorbed by cells.

Choosing an appropriate dynamic range for acquiring radio-frequency (RF) data from a high-frequency ultrasound (HFU) system is challenging because signals can vary greatly in amplitude because of focusing and attenuation effects. In addition, quantitative ultrasound (QUS) methods are sensitive to saturated data. In this study, the effects of saturation on QUS estimates of effective scatterer diameter (ESD) and effective acoustic concentration (EAC) were quantified using data acquired from 69 dissected human lymph nodes with a single-element transducer operating at a center frequency of 26 MHz. Artificially saturated signals (x(c)) were produced by thresholding the original, unsaturated RF echo signals (x). Saturation degree was quantified using Saturate-SNR. ESD and EAC were estimated from x(c) over a wide range of Saturate-SNR values. The value of the ESD estimate was minimally affected when Saturate-SNR ranged from infinity to 3.99 dB. However, the value of the EAC estimate decreased significantly with decreasing Saturate-SNR. EAC estimates were corrected using a linear relationship between EAC values over a range of Saturate-SNR values and l(1)-norm of x (i.e., the sum of absolute values of the true RF echo signal). To correct the estimates for ESD and EAC of saturated signals, the estimated original RF signal was derived from x(c) by a cubic spline. A line-arregression line was computed for the EAC values of a set of Saturate-SNR values vs. l(1)-norm. The intersection of the regression line and l(1)-norm of the estimated original signal gave a corrected value for EAC. The maximum error in the corrected EAC estimate was 0.48 dB/mm(3). The average differences in corrected ESD and EAC estimates of non-metastatic and metastatic lymph nodes were 7.80 mu m and 4.18 dB/mm(3), respectively. The error in the corrected EAC value was much smaller than the difference in the corrected values of non-metastatic and metastatic lymph nodes.

The frequency dependence of shear wave velocity provides significant information for evaluating viscoelastic character of tissue relating to liver fibrosis. Although Voigt model has been often used in viscoelastic analysis, several studies showed that the frequency dependence measured by dynamic mechanical analysis (DMA) test was not consistent with the theoretical prediction. To experimentally investigate the relationships of the change of the tissue structure and the viscoelasticity of tissue, the shear wave velocity of fatty and fibrotic livers of rat model was quantitatively measured by using shear wave elastography (SWE) and DMA test. In DMA test, shear wave velocity was calculated from the complex elasticity modulus; storage and loss elastic modulus. The difference in shear wave velocity between fatty and fibrotic livers was evaluated to be 0.27 m/s in SWE and 0.20 m/s in DMA test.

肝臓の繊維化に伴いミクロな領域で生じる組織性状変化を調べる目的で,肝線維症モデルのラット肝臓を対象に,超音波顕微鏡を用いて音速,減衰,音響インピーダンスの計測を行った.計測用に厚みが約10μmの薄切試料を作成し,中心周波数80MHz(方位分解能20μm)及び250MHz(方位分解能4μm)の超音波を用いて計測した.計測結果より求められるエコー信号の周波数スペクトルと理論的な計算結果をフィッティングすることで,音速,減衰,音響インピーダンスを推定した.線維組織の音速,減衰は他の部位や正常肝臓の値と比べて大きいことを確認した.一方,音響インピーダンスに関しては線維部分とその他部分で顕著な違いはみうけられなかった.

The development of a quantitative diagnostic method for organs using ultrasound would be highly medically significant. Detection and classification of tissue disease using the characteristics of the ultrasound echo signal, such as power spectrum, texture parameters, local attenuation and statistical characteristics, requires an understanding of the relationship between complicated scatterer properties and the echo signal. We developed a quantitative ultrasound (QUS) method for detecting and classifying liver fibrosis on the basis of the estimation of scatterer density from the statistical analysis of echo envelopes. This method and the other technique that estimate scatterer size in tissue were applied to the lymph nodes comprehensively for determining cancer metastasis.