透過您的圖書館登入
IP:18.116.19.17
  • 學位論文

應用於數位化乳房攝影之左右乳房自動校正電腦介面研發

Development of a Computer Interface for Automatic Alignment of Left and Right Breasts in Digitized Mammograms

指導教授 : 張元翔

摘要


在比較多重視角乳房攝影中之組織資訊時,乳房攝影之校正是一項重要的技術。本論文提出一應用於數位化乳房攝影之左右乳房自動校正電腦介面。研究方法包括前處理、自動特徵擷取(即頂點Top-Point與乳房中心軸Breast-Central-Axis)、及幾何運算(即位移與旋轉)之雙邊校正。本研究運用灰階聯合分配(Joint Gray Level Distribution)、相互資訊(Mutual Information)、與相關係數(Correlation Coefficient)以測量雙邊校正之性能。本論文結果乃是利用Mammographic Image Analysis Society(MIAS)數位化乳房攝影資料庫用進行評估。藉由50組乳房攝影所獲得的平均測量數據,校正前後之相互資訊測量值由0.9696提升至0.9476、相關係數亦由0.6519提升至0.8698。總結而言,本論文之自動乳房雙邊較正之電腦介面可用來協助放射科專家進行乳房攝影之判讀。另外,本技術亦可用來設計雙邊乳房攝影電腦輔助診斷系統之前處理步驟。

並列摘要


Mammogram alignment is an important technique when breast tissue information from multiple views of the breast is compared. This paper presents a computer interface for automatic alignment of left and right breasts in digitized mammograms. The methods include preprocessing, automatic detection of features (i.e., top-point and breast-central-axis), and bilateral alignment using geometric operations (i.e., shift and/or rotation). The performance of the bilateral alignment is quantitatively measured using the joint gray level distribution, the mutual information (MI), and the correlation coefficient. Our results were evaluated by the Mammographic Image Analysis Society (MIAS) Digital Mammogram Database. On the average of 50 mammogram pairs, the MI measures improved from 0.9696 to 0.9476 before and after alignment. The correlation coefficients are also improved from 0.6519 to 0.8698. Ultimately, our computer interface could be used to automatically align bilateral mammograms for radiologists during mammography interpretation. In addition, our techniques could be used a preprocessing in the design of computer-aided diagnosis (CAD) systems for the analysis of bilateral mammograms.

參考文獻


[1] World Health Organization, W.H.O. http://www.who.int/en/.
[2] Huo, Z., Giger, M., Vyborny, C. and Metz, C., "Breast cancer: Effectiveness of Computer-Aided Diagnosis Observer Study with Independent Database of Mammograms," Radiology 224(2), 560-568 (2002).
[3] Tourassi, G.D., Voracek, R.V., Catarious, D.M. and Floyd, C.E. "Computer-Assisted Detection of Mammographic Massed: A Template Matching Scheme Based on Mutual Information," Medical Physics 30, 2123-2130 (2003).
[4] Gur, D., "Computer-Aided Detection Performance in Mammographic Examinations of Masses: Assessment," Radiology, 233, 418-423 (2004).
[5] Catarious, D.M., Baydush, A.h. and Floyd C.E., "Incorporating of An Iterative, Linear Segmentation Routine into A Mammographic Mass CAD System," Medical Physics 31(6), 1512-1520 (2004).

延伸閱讀