非酒精性脂肪性肝病是現今最常見的肝臟代謝疾病。許多研究證明了代謝性肝病、肝癌、心血管疾病以及糖尿病有著密切的關係。超音波成像系統是目前臨床上最常用來檢測非酒精性脂肪肝的診斷儀器，因為其具有即時成像的能力、非侵入式的診斷方式、安全無害、低成本以及可攜帶式的機身。然而，目前超音波儀器的診斷方式都是透過人眼去判讀超音波影像，此種診斷方式取決於操作者的經驗以及醫師的主觀判斷，因此有可能會使診斷結果有誤。為了改善診斷脂肪肝病時的視覺影響以及高度的主觀判斷，便有人提出了以定量參數的分析方法用於診斷非酒精性脂肪肝。
在本研究中，我們因應醫師於臨床上診斷脂肪肝時所注意的特徵並提出了以估測自相依函數的半高全寬、平均散射子距離、Nakagami分布、K分布，以及影像強度衰減來做為定量參數分析。我們將241位受測者的樣本資料分別分成兩類(沒有脂肪肝以及有脂肪肝)和三類(沒有脂肪肝、輕度脂肪肝、中度至重度脂肪肝)並觀察這些定量參數是否對脂肪肝的分類是有用的。從結果得知，分成兩類時，其接收者操作特徵曲線的曲線下面積為0.8105，而分成三類時，其曲線下面積分別為0.843、0.733以及0.868。這些曲線下面積都遠超過於0.5，因此我們認為這些參數定量分析方法，對於非酒精性脂肪肝的診斷的確是有幫助的。

Non-alcoholic fatty liver disease (NAFLD) is the most common liver metabolic disease nowadays. Many researches demonstrate that the metabolic liver disease has close relationship with liver cirrhosis, hepatocellular, cardiovascular disease and diabetes. Ultrasound imaging system is a first diagnostic tool to detect NAFLD because of its real time, non-invasive, safety, low cost and portable. However, a human visual inspection of the ultrasound images is not accurate enough to diagnose the NAFLD stage and the diagnostic result depends on doctor’s subjective operator. To improve the visual and highly subjective criteria for diagnosing fatty liver disease, quantitative imaging manners have been proposed.
In this study, we observed the features of clinical diagnosis, which are used for classifying the NAFLD by clinical experts, and then we proposed the corresponding quantitative parameters, the full width half maximum estimation of auto-covariance function, mean scatterer spacing, Nakagami distribution, K distribution and intensity attenuation as the quantitative parameters for analysis. These data, which are the liver parenchyma views of 241 subjects, were classified into 2 categories (normal liver and fatty liver) and 3 categories (normal, mild and moderate to severe) to observe if these quantitative features are useful for classification. The results show the area under ROC curve of 2 categories and 3 categories classified by using all features are 0.8105, 0.843, 0.733 and 0.868 for 2 categories and 3 categories, respectively. These values exceed 0.5; hence, we think these quantitative features are useful for classification.

[1] “105年國人死因統計結果-衛生福利部全球網站” http://www.mohw.gov.tw/cp-16-33598-1.html
[2]Mayo Clinic Staff, “Nonalcoholic fatty liver disease,” http://www.mayoclinic.org/diseases-conditions/nonalcoholic-fatty-liver-disease/home/ovc-20211638
[3]Mary E. Rinella, MD “Nonalcoholic Fatty Liver Disease A Systematic Review,”American Medical Association, vol. 313, no. 22, pp. 2263-2273, 2015
[4]中國醫藥大學醫學系解剖科,中國醫藥大學附設醫院放射部“脂肪肝|醫學影像園地” http://www2.cmu.edu.tw/~cmcmd/ctanatomy/clinical/clinical/fattyliver.html
[5]Y. Y. Liao, K. C. Yang, “Multifeature analysis of an ultrasound quantitative diagnostic index for classifying nonalcoholic fatty liver disease,” Scientific Reports, vol. 6, 2016.
[6]P. H. TSUI, C. C. CHANG, “Imaging local scatterer concentrations by the Nakagami statistical model,”Ultrasound in Med. & Biol., vol. 33, no. 4, pp. 608-619, 2007
[7]S. Ballestri, A. Lonardo, “Ultrasonographic fatty liver indicator, a novel score which rules out NASH and is correlated with metabolic parameters in NAFLD,” Liver International, vol.32, no. 8, pp. 1242-1252, 2012
[8]“Ultrasound of the liver,” http://www.ultrasoundpaedia.com/normal-liver2/
[9]M. L. Oelze, J. Mamou, “Review of Quantitative Ultrasound: Envelope Statistics and Backscatter Coefficient Imaging and Contributions to Diagnostic Ultrasound,” IEEE Trans. Ultrason. Ferroelectr. Freq. Control, vol. 63, no. 2, 2016
[10]P. Bharti, D. Mittal, “Computer-Aided Characterization and Diagnosis of Diffuse Liver Diseases Based on Ultrasound Imaging: A Review,” Ultrasonic Imaging, vol. 39, no. 1, pp. 33-61, 2016
[11]G. WEIJERS, G. WANTEN, “Quantitative Ultrasound for Staging of Hepatic Steatosis in Patients on Home Parenteral Nutrition Validated with Magnetic Resonance Spectroscopy: A Feasibility Study,” Utrasound in Med. & Biol., vol. 42, no. 3, pp. 637-644, 2016
[12]X. Qu, T. Azuma, “Average sound speed estimation using speckle analysis of medical ultrasound data.,” Int. J. CARS., vol. 7, no. 6, pp. 891-899, 2012
[13]Robert F. Wanger, Stephen W. Smith, “Statistics of Speckle in Ultrasound B-Scans,” IEEE Trans. Sonics Ultrason., vol. 30, no. 3, pp. 156-163, 1983
[14]Keith A.Wear, Robert F. Wagner, “Application of autoregressive spectral analysis to cepstral estimation of mean scatterer spacing,” IEEE Trans. Ultrason. Ferroelectr. Freq. Control, vol. 40, no. 1, pp. 50-58, 1993
[15]Udantha R. Abeyratne, Xiaoyan Tang, “Ultrasound scatter-spacing based diagnosis of focal diseases of the liver,” Biolmedical Signal Processing and Control, vol. 2, no.1, pp. 9-15, 2007
[16]Linda L. Fellingham, F. Graham Sommer, “Ultrasonic Characterization of Tissue Structure in the In Vivo Human Liver and Spleen,” IEEE Trans. Sonics Ultrason., vol.31, no. 4, pp. 418-428, 1984
[17]Christoph B. Burckhardt, “Speckle in ultrasound B-mode scans.,” IEEE Trans. Sonics Ultrason., vol. 25, no. 1, pp. 1-6, 1978
[18]P. H. Tsui, C. C. Chang, “Imaging local scatterer concentrations by the Nakagami statistical model.,” Utrasound in Med. & Biol., vol. 33, no. 4, pp. 608-619, 2007
[19]F. DESTREMPES, G. CLOUTIER, “A critical review and uniformized representation of statistical distributions modeling the ultrasound echo envelope.,” Utrasound in Med. & Biol., vol.36, no. 7, pp. 1037-1051, 2010
[20]Li Weng, John M. Reid, P. Mohana Shankar, “Ultrasound speckle analysis based on the K distribution.,”J. Acoust. Soc. Amer., vol. 89, no. 6, pp. 2992-2995, 1991
[21]J. W. Goodman, “Statistical properties of laser speckle patterns,” in Laser Speckle and Related Phenomena, J. C. Dainty, Ed. Berlin: Springer-Verlag, vol. 9, pp. 9-75, 1975
[22]C. C. Chang, C. J. Lin, “LIBSVM -- A Library for Support Vector Machines,” https://www.csie.ntu.edu.tw/~cjlin/libsvm/
[23]M. R. Li, “Diagnosis of fatty liver disease by ultrasound multi-characteristic parameters,” 103.