使用超音波力學散射統計參數影像指標定量肝纖維化程度

Abstract

本研究致力於探討超音波影像於臨床肝纖維化診斷的應用，主要針對超音波RF訊號轉換後的包絡線訊號，探究其正常組織與病變組織統計分佈上的差異。肝硬化幾乎是一個不可逆的過程，且目前尚無有效的方法對已經硬化的肝進行復原。肝硬化形成的機制是肝細胞受到傷害後引發一系列慢性的、長期性的退化，結果導致肝臟內部結構重組。通常肝纖維化與脂肪肝的演發，都有可能促使肝硬化的加劇。侵入式診斷例如肝穿刺切片，會造成病人的痛苦且帶有些微機率的後遺症。所以現在臨床上希望建立起一套有效且可靠的非侵入式評分方法，期望能預測到早期的肝硬化現象。 超音波B-Mode影像存在斑紋現象，屬於一種逆散射訊號隨機干涉造成的結果。文獻中提到，這種斑紋現象在肝臟纖維化或脂肪肝的情況中，且具有某種程度的鑑別程度。斑紋現象的RF訊號特性，與該組織散射粒子密度有一定關係，且斑紋所呈現的樣式在健康與疾病組織中也會有所差異。三種分析方法在本研究中被使用：Nakagami-m參數分析、Yamaguchi方法分析、ACRA(adaptive criteria-referenced assessment)方法分析。Nakagami-m參數分析使用m值來描述分析區域的訊號分佈情形(m<1: pre-Rayleigh, m=1: Rayleigh, m>1: post- Rayleigh)，從文獻中已經證實對乳房腫瘤與眼球白內障有著很好的判斷效果。Yamaguchi方法分析使用仿體實驗得到的結果所做的參數假設，去判斷該分析區域是否為纖維化組織。ACRA方法則是本實驗室的原創方法，其特點是能適應各種不同的實驗條件，與自身正常組的標準常模進行差異比較，得到的差異程度可有效反映到肝臟纖維化的病變。 實驗結果顯示，在老鼠肝臟離體實驗中，Nakagami-m參數與ACRA方法在老鼠肝臟離體實驗有效地將正常組與纖維化組老鼠區分開來。本研究利用接受者操作特徵曲線下面積(AUC)來判別評分效果的好壞。在臨床線性陣列探頭實驗中，三種方法在初期纖維化F≥1分類中的AUC表現良好(Nakagami-m: 0.86、Yamaguchi: 0.98、ACRA: 0.95)。在臨床弧形陣列探頭實驗中，三種方法在F≥1與F≥2分類中的AUC表現良好(Nakagami-m: 0.96, 0.82、Yamaguchi: 0.93, 0.92、ACRA: 0.99, 0.93)。臨床術後肝臟離體掃描實驗中，只有ACRA方法能有效區分出F≥2與F=4的分類情形(0.87、0.93)。ACRA方法的運作概念還可以延伸到其他的領域，它能適應不同的實驗條件，透過訊號區域性的統計特性，觀察其正常與病變組織的差異程度，反映到病變組織的異常程度，此方法非常具有潛力以及臨床應用價值。

This study is dedicated to the application of liver fibrosis assessment using ultrasound imaging, focusing on the envelope signal converted from RF signal to investigate the differences of the statistical distributions between healthy and disease livers. Liver cirrhosis is considered as an irreversible process that there is almost no effective course of treatment to recover the fibrosis liver. Fibrosis is marked by the gradual replacement of hepatocytes by extracellular collagen, which is a chronic and long-term degeneration of the liver function. Invasive assessment like liver biopsy often accompanied with sequela and is prone to sampling error when small biopsy samples are analyzed. Therefore, there is a clinical need to establish an effective and reliable method for the noninvasive assessment, espesically the prediction of early stage liver fibrosis. Ultrasound speckle in B-mode images, which is the result of a wave interference phenomenon of backscattering signal, has been used for clinical liver fibrosis and steatosis diagnosis. The characteristic of speckle is associated with the density of scatterers in tissue, indicating different patterns among various tissue properties. To analyze the RF signal derived from liver ultrasound, three methods were implemented: Nakagami-m parameter method, Yamaguchi method and ACRA (adaptive criteria-referenced assessment) method. Nakagami-m parameter method uses m-value to describe the distribution within the ROI (region of interest), proved to be a valid approach characterizing breast tumors and cataract lens. Yamaguchi method determines whether the ROI is a fiber area or not based on the assumptive parameters derived from the phantom experiment. ACRA is a method, which was originally developed by our lab and could be adapted to various system conditions; it reflects the degree of fibrosis liver via the unassessed tissues in comparison with a criteria-reference constructed by healthy tissues. Results showed that Nakagami-m and ACRA method were able to discriminate the degree of fibrosis from livers in vitro rat experiment. Area under receiver operating characteristic curve (AUC) was used to judge the performence. In the in vivo linear array experiment, three methods performed well in the early stage F≥1 fibrosis (Nakagami-m: 0.86, Yamaguchi: 0.98, ACRA: 0.95). In the in vivo convex array experiment, three methods performed well in the stage F≥1 and F≥2 fibrosis (Nakagami-m: 0.96, 0.82, Yamaguchi: 0.93, 0.92, ACRA: 0.99, 0.93). In the in vitro operation linear array experiment, only ACRA successfully identified stage F≥2 and F=4. It is concluded that the concept of ACRA could exend to other domains, observing the degree of the difference between tissues to indicate the abnormity. This method possesses clinical potential and application value.