使用雙重Nakagami統計模型進行肝實質病變之超音波評估

Ultrasound evaluation of liver disease using the double Nakagami distribution

陳盈宏(Ying-Hung Chen)

指導教授 ： 張建成

共同指導教授 ： 崔博翔(Po-Hsiang Tsui)

國立臺灣大學/工學院/應用力學研究所/碩士(2020年)

https://doi.org/10.6342/NTU202003462

摘要

肝臟是沈默的器官，身體發出警訊的時候往往已事態嚴重，且目前的黃金標準-病理切片存在著侵入式診斷的一些缺點，在早期的肝硬化的測量上不是會優先採取的方式，而超音波檢測有即時成像及非侵入式的優點成為臨床醫學上重要的檢測工具。定量式超音波信號診斷方法相比於傳統的灰階影像更能反映不同組織散射子的物理特性，以導入參數成像的方法並分析輸出參數與肝實質病變程度的關係，達到評斷病變程度的目的。
本研究使用雙重Nakagami統計模型分別分析超音波逆散射訊號的隨機干涉，針對病變組織與健康肝細胞的散射訊號振幅分佈差異做出區分，再以不同的參數輸出做成參數影像。Double Nakagami有別於單Nakagami將所有訊號視為一體，將脂滴及正常肝細胞分開計算，再以EM及FM算法找出各分佈的參數做疊合，最後尋找合適的參數輸出與病理切片的分群結果做比較。
本實驗最終在評估脂肪肝中以EM_c算法配合輸出參數μ_F在辨別輕度、中度及重度脂肪肝中分別達到AUC值0.77、0.84、0.84。在高纖維化病人評估脂肪肝的方面以EM算法配合輸出參數μ_F在辨別輕度、中度及重度脂肪肝中分別達到AUC值0.77、0.85、0.84。在評估肝纖維化方面以LC model結合四種參數配合EM_c算法在辨別F1、F2、F3、F4達到AUC值0.71、0.63、0.59、0.64。最後在高脂肪肝病人評估肝纖維化以LC model結合四種參數配合FM_c算法在辨別F1、F2、F3、F4達到AUC值0.93、0.77、0.73、0.81。

關鍵字

脂肪肝 ； 肝纖維化 ； 超音波參數影像 ； Double Nakagami ； Linear Combination

並列摘要

The liver is known as a silent organ. When the notable symptoms occur, it turns out to be too late. There are some disadvantages of the current gold standard -liver biopsy. Its’ invasive diagnosis not a preferred method for the evaluation of the early liver cirrhosis. On the contrary, the real-time and non-invasive characteristics of ultrasound imaging made it becomes an important tool in clinical medicine. Compared with the traditional B-mode imaging, the quantitative ultrasonic signal diagnosis method can show the physical characteristics of different backscattered signals. To determine the extent of lesions, this study analyze the relation between the ultrasonic parameters and liver disease by introducing the method of parametric imaging.
Double Nakagami statistical model was used in this study to analyze backscattered signals, distinguish the difference of amplitudes between the backscattered signals of lipid droplets and normal Hepatocytes and make parametric images by output parameters. As opposed to Nakagami consider all signals as one, double Nakagami takes lipid droplets and normal Hepatocytes apart. After that EM and FM algorithms were used to find out the parameters of each distribution. Comparisons are made between the results of liver biopsy.
The results indicate the AUC values of 0.77, 0.84, and 0.84 correspond to the mild, moderate, and severe fatty liver by evaluating the EM_c algorithm and the output parameter for all patients. Upon evaluating the extent to fatty liver of patients with high liver fibrosis, the AUC values of 0.77, 0.85 and 0.84 correspond to the mild, moderate, and severe fatty liver. The evaluation of liver fibrosis for all patients that based on the LC model using four parameters with the EM_c algorithm, resulted in the AUC values of 0.71、0.63、0.59、0.64 correspond to the stage F1、F2、F3、F4. Finally, the evaluation of liver fibrosis with high fatty liver patients that based on the LC model using four parameters with the EM_c algorithm, resulted in the AUC values of 0.93、0.77、0.73、0.81 correspond to the stage F1、F2、F3、F4.

並列關鍵字

fatty liver ； liver fibrosis ； parametric imaging ； double Nakagami ； linear combination

參考文獻

[1] P.-H. Tsui, Z. Zhou, Y.-H. Lin, C.-M. Hung, S.-J. Chung, and Y.-L. Wan, "Effect of ultrasound frequency on the Nakagami statistics of human liver tissues," PloS one, vol. 12, no. 8, p. e0181789, 2017.

[2] C. B. Burckhardt, "Speckle in ultrasound B-mode scans," IEEE Transactions on Sonics and ultrasonics, vol. 25, no. 1, pp. 1-6, 1978.

[3] P. M. Shankar, "A general statistical model for ultrasonic backscattering from tissues," IEEE transactions on ultrasonics, ferroelectrics, and frequency control, vol. 47, no. 3, pp. 727-736, 2000.

[4] P. Shankar, "Statistical modeling of scattering from biological media," The Journal of the Acoustical Society of America, vol. 111, no. 5, pp. 2463-2463, 2002.

[5] P.-H. Tsui and C.-C. Chang, "Imaging local scatterer concentrations by the Nakagami statistical model," Ultrasound in medicine biology, vol. 33, no. 4, pp. 608-619, 2007.