3-D乳房剪力波彈性影像之腫瘤診斷

Abstract

乳癌一直是全球女性的十大死因之一，而腫瘤的軟硬程度也已經被證實為分辨良惡性的主要特徵。不同於傳統的乳房彈性超音波，在本實驗中所使用的乳房剪力波彈性超音波只需利用聲波輻射便可取得腫瘤的彈性硬度。本實驗更使用了能夠提供完整組織彈性資訊的三維乳房剪力彈性超音波。首先，我們將三維剪力波的多張影像合成一張新的影像，此張新的影像只保留各張圖片中最硬的資訊。接著我們基於模糊平均分群法來去除一些合成影像過程所產生的一些較多餘得彈性資訊。而我們論文的目的即是針對此影像來擷取出彈性特徵並用以診斷腫瘤的良惡性。最後，我們會比較二維以及三維剪力波影像在診斷上的差異。本實驗的病例共有51個，其中包含25個良性與26個惡性的病例。根據實驗結果，本篇論文提出的方法對於三維及一般二維剪力波影像作腫瘤診斷的準確度分別為94.12%及80.39%。經由實驗結果分析，三維影像在腫瘤分辨的準確度上會有顯著的提升。

The breast cancer is always one of the top ten death causes for women around the world. The firmness of the tumors has been proved to be an important characteristic for differentiating benign and malignant tumors. Different from the conventional sonoelastography, this paper employs the shear wave elastography that uses the acoustic radiation substituting the manual tissue compression to generate the information of tumor firmness. The new three-dimensional (3-D) shear wave elastography technique that provides more complete of data on tissue firmness of the mass is also adopted in this study. We aim to merge the eight elastographic slices from a 3-D image into one image at first. We then reduce the insignificant elasticity information of the merged image based on fuzzy c-means clustering and extract the elasticity information to diagnose tumors. Finally, the diagnostic performances of 2-D and 3-D shear wave elastography for tumor diagnosis are compared. In this study, we use 51 breast tumors composed of 25 benign and 26 malignant cases. The experimental results of 3-D and 2-D shear wave elastography illustrate that the accuracy in distinguishing tumors are 94.12% and 80.39%, respectively. Based on statistical analyses of experimental results, the diagnostic performances on the 3-D images are significantly better than those of 2-D images.