以乳房動態對比增強核磁影像之血管分析與擴散加權核磁影像之表面擴散係數基礎的乳癌腫瘤輔助診斷

Abstract

近年的研究顯示，新生血管的癥狀在乳癌的預後中扮演著重要的角色。另外，鄰近乳房腫瘤血管的巨增也常被用來檢測惡性腫瘤的一個現象。因此，鄰近腫瘤周圍的血管能夠提供有效的資訊並且使用在乳癌良惡性診斷之中。本篇論文藉由電腦的輔助去綜合分析動態對比增強的核磁共振成像 (DCE-MRI)以及擴散加權影像 (DWI)所提供的一些腫瘤內部常用特徵，並且提出一個量化的方法去檢測腫瘤周圍的血管，進一步地去評估腫瘤周圍的血管資訊是否能提供更好的診斷結果。本篇論文實驗包含50個良性及54個惡性的乳癌腫瘤病例，根據實驗結果顯示，結合腫瘤外部的鄰近血管以及其他常用的腫瘤內部特徵能有效地區分腫瘤的良惡性並且提高診斷的結果，能達到準確性95.2%、敏感性96.3%、專一性94.0%以及Az值0.95。

Angiogenesis sign plays a crucial role in breast tumor prognosis and the increased vascularity adjacent to breast tumor is reported to be useful in the malignancy evaluation. In this study, a quantitative method was proposed to extract the vessel information surrounding the tumor area in dynamic contrast enhanced magnetic resonance imaging (DCE-MRI). First, a ring-shaped region encompassed a tumor was segmented. The adjacent vessels were then extracted from the ring-shaped region. The quantitative vessel features were combined with the shape, texture, and apparent diffusion coefficient (ADC) extracted from the tumor area to improve the diagnostic performance of the proposed computer-aided (CAD) system. The collected cases used in the experiment included 50 benign and 54 malignant cases. As a result, the proposed CAD system using the combination of all features achieved the accuracy, sensitivity, specificity, and Az value of 95.2% (99/104), 96.3% (52/54), 94.0% (47/50), and 0.95, respectively. The accuracy was significantly better than that of individual kind of features (p-value<0.05). In summary, the proposed CAD system based on the quantitative vessel features would be promising in the diagnosis of breast cancer.