Please use this identifier to cite or link to this item:
標題: 分水嶺分割法於子宮頸抹片細胞影像之分析
The Analysis of Cervical Smear Images based on Watershed Segmentation
作者: 林宛怡
Lin, Wan-Yi
關鍵字: watershed segmentation
region merging algorithm
cervical smear image
出版社: 電機工程學系所
引用: [1] 楊辰夫,李淑美 譯註,“細胞診斷學 Practical cytodiagnosis”,合記圖書,1992年。 [2] 張智星 著,“MATLAB 程式設計與應用”,清蔚科技,2000年。 [3] 穆紹綱 編著,“數位影像處理-活用Matlab”,全華科技,2003年。 [4] 黃博琪,“子宮頸細胞自動化電腦輔助影像分析系統之設計研發”,國立中興大學,碩士論文(92) 。 [5] 馮齡儀,“電腦輔助子宮頸抹片異常細胞辨識之初期研究”,私立中原大學,碩士論文(93) 。 [6] 陳淑嬌,“梯度向量流主動輪廓模型於子宮頸抹片細胞邊界偵測之應用”,國立中興大學,碩士論文(91) 。 [7] 行政院衛生署, [8] 行政院衛生署台中醫院病理科, [9] 馬偕紀念醫院病理科, [10] R. C. Gonzalez and R. E. Woods, “Digital Image Processing,” second edition, Prentice Hall, 2002. [11] A. P. Dhawan, “Medical Image Analysis,” Wiley-IEEE Press, 2003. [12] H. Fu, Z. Chi, D. Feng, “Attention-driven image interpretation with application to image retrieval,” Pattern Recognition, vol. 39, no. 9, pp. 1604 – 1621. [13] Z. Hou, Q. Hu, W.L. Nowinski, “On minimum variance thresholding,” Pattern Recognition Letters, vol. 27, no. 14, pp. 1732 – 1743. [14] Women''s Health & Education Center, [15] ThinPrep System,
摘要: 近年來利用工程技術解決醫學問題的相關應用,已有相當顯著之進展。其中,數位影像處理的技術,對於分析醫學影像的範疇,更有長足之助益。在子宮頸癌的臨床診斷中,新型柏氏超薄子宮頸抹片始終扮演著關鍵的角色。然而繁瑣的人工閱片過程,卻極易導致人為疏失或偽陰性的檢測結果。為了協助病理師檢驗抹片樣本,我們提出以分水嶺演算法為基礎的影像處理技術,利用分水嶺分割法使目標診斷區域更易觀測,以期提高閱片的效率。 本研究根據形態學分水嶺之概念所提出的方法,使分割結果更趨於穩定,同時也包含了連續的分割邊界。除此之外,以子宮頸抹片細胞之特徵參數為基礎的區塊合併演算法,也解決了過度分割的問題。由於此方法能辨識出抹片影像中肉眼不易獲得的資訊,因此能有效提高檢驗的準確性以及臨床診斷的正確度。
In recent years, the application of engineering techniques to solve the problems in medical field has been under rapid developments. Among which, the digital image processing is particularly applicable to medical image analysis. In the diagnosis of cervical cancer, the Thin-Prep Pap Smear has always played a key role. However, the tedious primary screening is prone to bring about human errors and false negative test results. In order to assist pathologists in examining the Thin-Prep Pap Test specimens, a new method based on the watershed algorithm is proposed to improve the efficiency of the screening process. The proposed approach is based on the concept of morphological watersheds and it can produce stable segmentation results. Furthermore, with a merging algorithm using the characteristics of cervical smear images, the problem of over-segmentation is alleviated as well. With the capability of extracting the required information effectively, the proposed method has the potential to achieve high accuracy in cervical screening.
其他識別: U0005-2407200700442500
Appears in Collections:電機工程學系所



Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.