乳房腫瘤超音波特徵之量化與效力分析

Abstract

超音波檢查為早期乳房腫瘤最有效的非侵入性篩檢方法之一，專業醫師根據所觀察之影像特徵提出持續追蹤或進一步檢查之建議。由於目前超音波影像皆為經由醫師主觀擷取並判斷，不同觀察者對於相同影像常有不同解讀且差異明顯的檢查結果，影像特性之客觀量化因此成為重要課題。 本研究以超音波於乳房腫瘤影像特性之量化指標為主要研究焦點，根據乳房腫瘤於超音波影像上的臨床特性，分別提出與建立其於超音波影像之量化指標，其中於灰階超音波影像上包含異質性指標、輪廓特色指標及後方回音陰影指標，而於彈性超音波影像上則提出探究腫瘤內、腫瘤邊緣以及其周圍組織的硬度特性指標，並透過醫學研究中常用於診斷腫瘤良惡性靈敏度的方法─接收者操作特徵曲線(Receiver Operative Characteristic Curve, ROC)，針對每一指標對於乳房腫瘤良惡性之判斷績效做靈敏度的驗證。 本研究利用臺大醫院所提供的264筆乳房腫瘤的樣本資料來進行特徵之量化。最後，本研究將表現顯著的量化指標透過費雪判別分析尋找一最佳線性組合，期望所提出的顯著量化指標能使乳房腫瘤的良惡性判別最佳化，並以臺大醫院所提供的病歷良惡性診斷資料作為判別準確性的評斷，最佳結果可得AUC達0.8957。

Ultrasound (US) imaging is one of the most effective non-invasive screening tools for tumors of early stage. Based on observation impressions of US images, clinicians make suggestions for patients to be subject to periodic follow-up or further cytologic tests. Because acquisition and observation of ultrasound images are mostly subjective and highly dependent on the medical staff’s experience and judgment, the observer variation often results in significantly different decisions. Objective quantification of sonographic tumor features has become a pressing issue facing the medical staff.Hence, this research focus on the quantitative indices of sonographic breast tumor features. Base on the clinical research, we are developed including texture heterogeneity indices, morphologic indices and posterior acoustic shadow indices in gray-scale image. In elastography image, we are developed elastographic indices within the tumor, its margin and its adjacent tissues. The newly quantitative indices are further validated through their performance of the receiver operating characteristic (ROC) curves in screening and prognosis of the breast cancer. To validate the performance, we use a database of 264 cases (65 malignant lesions and 199 benign solid lesions) provided by National Taiwan University Hospital (NTUH) to retrieve these quantitative indices. Furthermore, the Fisher Linear Discriminant is employed on the significant quantitative indices to obtain a linear combination for a better classification power. The clinical report is then used to evaluate the diagnosis Accuracy. The best combination of significant quantitative indices can get the result of AUC achieve 0.8957.