臺灣博碩士論文加值系統

陰道鏡是子宮頸癌篩檢中非常重要的工具，醫師可以透過它觀察子宮頸上皮組織的實際情況判斷病變等級。由於病變情況沒有客觀的量化標準，所以必須完全依靠醫師的主觀判斷。若是醫師經驗不足，則準確度會降低。我們可以運用電腦影像技術去輔助醫師確認病灶位置，並且提供病灶癌化程度的量化數據與分級結果，藉此讓醫師進行更精準的診斷。本研究可以自動化切割出病灶位置，並且將癌化特徵抽取出來進行分級。病灶的切割過程使用Fuzzy k-mean clustering、自我組織圖網路和分群品質測量進行影像切割。接著抽取病灶的特徵向量，我們使用HSV color model取得病徵的特徵向量。針對特徵向量使用SVM、KNN分類器進行分級。我們使用白化特徵進行分級，正確分級成功率達87.2%。我們相信自動化的輔助診斷系統，能夠幫助醫師進行診斷及預後工作。

The morbidity and mortality of cervical cancer can be reduced by the screening of the precancerous lesions. Pap smears, colposcopy and biopsy are the most common screening tools. Pap smear is the first-line tool because of its high specificity and low cost. But its false-positive rate is too high and must be confirmed by other tools. Biopsy is a deterministic examination for cervical neoplasia. However, it is not suitable for the high probability of false-positive. Digital colposcopy is a promising technology for the detection of cervical intraepithelial neoplasia. However, there are no quantitative criteria for the differential of precancerous lesions and it is subjected to the variation of inter-observer and intra-observer. Therefore, automated image analysis of colposcopic images is thus necessary for the improvement of diagnosis of colposcopy. The segmenetation of the lession from digital colposcopyic image is a key issue of this analysis. Our goal is to develop a segmentation policy that can separate images into regions which contain the lesion areas. These areas can be provided to the analysis system and help doctor to make the diagnosis.

目 錄
摘 要 I
Abstract II
圖 目 錄 V
表 目 錄 VI
第一章 緒論 1
1.1研究動機 1
1.2研究目的 3
1.3論文架構 3
第二章 文獻探討 4
2.1 子宮頸及子宮頸癌 4
2.2子宮頸癌的診斷 7
2.3陰道鏡檢查及子宮頸癌病灶的判斷 8
2.4 群聚及分群演算法 12
2.4.1 k-means clustering 13
2.4.2 Fuzzy k-means clustering 14
2.5自我組織圖網路 16
2.6 分群品質測量(Cluster-Quality Measures, CQM) 20
2.6.1 Global Silhouette Index(GSI) 21
2.6.2 Partition Index(PI) 21
2.6.3 Separation Index (SI) 22
2.6.4 Dunn’s Index(DI) 22
2.8 HSV color model 24
第三章 研究方法 25
3.1研究材料與設備 25
3.2 無監督式分割 25
3.3 癌化區域選擇 36
3.4 特徵抽取及分類 37
3.4.1白化程度特徵抽取 37
3.4.2分類器選擇(SVM、KNN): 37
第四章 實驗與討論 38
4.1 影像來源 38
4.2 實驗設備 38
4.3 實驗結果與討論 39
4.3.1 SOM神經元的數目 39
4.3.2 SOM對效率的影響 40
4.3.3白化程度特徵的分類效率 41
第五章 結論 43
參考文獻 44

圖 目 錄
圖一 子宮解剖圖 .............................................................................................. 4
圖二 三種癌前病變的陰道鏡影像 .................................................................. 6
圖三 塗抹醋酸前後影像 .................................................................................. 9
圖四 SOM 映射方式 ...................................................................................... 16
圖五 六角形的鄰近範圍形狀 ........................................................................ 19
圖六 分群結果比較 ........................................................................................ 20
圖七 Feature space .......................................................................................... 23
圖八 HSV 色彩模型 ....................................................................................... 24
圖九 四種分群品質計算分法 ........................................................................ 28
圖十 各種群內距離計算的結果 .................................................................... 29
圖十一 分群數為4 的執行結果 .................................................................... 30
圖十二 系統流程圖 ........................................................................................ 35
圖十三 群組與中心位置示意圖 .................................................................... 36
圖十四 各等級癌化區域飽和度分佈情況 ..................................................... 42

表 目 錄
表一 四種分群品質計算方法的分析............................................................. 27
表二 不同神經元數目對correctness 的影響 ................................................. 39
表三 各個癌化等級的分群效率 .................................................................... 40

[1][online]世界最新癌訊(2003)，第十輯。
[2][online]Cervical Disease and Neoplasia(2009), Brookside Associates Medical Education Division website.
[3][online]Cancer Facts & Figures(2006), available at: http://www.cancer.org.
[4][online]Color model, http://zh.wikipedia.org/wiki/HSV.
[5][online]Harvard Health publications, www.health.harvard.edu/
[6]R. Ali, U. Ghani, and A. Saeed(1998), Data clustering and its applications. Available at: http://www.members.tripod.com/asim_saeed/paper.htm.
[7]A. M. Bensaid, L. O. Hall, J. C. Bezdek, L. P. Clarke, M. L. Silbiger, J. A. Arrington, and R. F. Murtagh(1996), “Validity-guided (Re) clustering with applications to image segmentation,” IEEE Trans. on Fuzzy Systems, 4(2), 112-123.
[8]C. J. Balas, G. C. Themelis, E. P. Prokopakis, I. Orfanudaki, E.Koumantakis, and E. S. Helidonis(1999), “In vivo detection and staging ofepithelial dysplasias and malignancies based on the quantitative assessmentof acetic acid-tissue interaction kinetics,” J. Photochem.Photobiol., B 53, 153–157.
[9]C. Balas(2001), “A novel optical imaging method for the early detection,quantitative grading, and mapping of cancerous and precancerouslesions of cervix,” IEEE Trans. Biomed. Eng. 48, 96–104.
[10]J. C. Bezdek and N. R. Pal(1988), “Some new indexes of cluster validity,” IEEE Trans. on Systems, Man and Cybernetics, part B 28(3), 301-315.
[11]N. Bolshakova, and F. Azuaje(2003), “Clustering validation techniques for genome expression data,” Genomic Signal Processing, 83 (4), 825-833.
[12]J. T. Cox, M. Schiffman, and D. Solomon(2003), “Prospective follow-upsuggests similar risk of subsequent cervical intraepithelial neoplasiagrade 2 or 3 among women with cervical intraepithelial neoplasiagrade 1 or negative colposcopy and directed biopsy,” Am. J. Obstet.Gynecol. 188, 1406–1412.
[13]J. C. Dunn(1974), “Well separated clusters and optimal fuzzy partitions,” J. Cybernetics, 4, 95-104.
[14]D. Ferris, J. T. Cox, D. M. O’Connor, V. C. Wright, and J. Foerster(2004), Modern Colposcopy: Textbook and Atlas, American Society for Colposcopy and Cervical Pathology, Dubuque, IA.
[15]R. E. Fan, K. W. Chang, C. J. Hsieh, X. R. Wang, and C. J. Lin(2008), “LIBLINEAR - A Library for Large Linear Classification,”, Machine Learning Research, 9, 1871-1874.
[16]S. Gordon, G. Zimmerman, R. Long, S. Antani, J. Jeronimo, and H.Greenspan(2006), “Content analysis of uterine cervix images: initial stepstowards content based indexing and retrieval of cervigrams,” Proc.SPIE 6144, 1549–1556.
[17]J. Han, M. Kamber, and A. K. H. Tung(2001), “Spatial clustering methods in data mining: A survey In H. Miller, & J. Han (Eds.),” Geographic data mining and knowledge discovery. Taylor and Francis.
[18]J. A. Hartigan and M. A. Wong(1979), “A K-Means Clustering Algorithm,” Applied Statistics, Vol. 28, No. 1, 100-108.
[19]H. Kaufman, A. Zelenchuk, R. Flewelling, P. Schmid, and Z. Hed(2005),“Methods of monitoring effects of chemical agents on a sample,” U.S. Patent No. 6, 902-935 B2.
[20]W. Li, S. Venkataraman, U. Gustafsson, J. C Oyama, D. G Ferris, R. W Lieberman(2009), “Using acetowhite opacity index for detecting cervical intraepithelial neoplasia,” j.Biomed. Opt. , 14(1), 014020.
[21]R. Nock, and F. Nielsen(2006), “On Weighting Clustering," IEEE Trans. on Pattern Analysis and Machine Inteligence, 28 (8), 1–13.
[22]B. W. Pogue, M. A. Mycek, and D. Harper(2000), “Image analysis fordiscrimination of cervical neoplasia,” J. Biomed. Opt. , 5, 72–82.
[23]B. W. Pogue, H. B. Kaufman, A. Zelenchuk, W. Harper, G. C. Burke,E. E. Burke, and D. M. Harper(2001), “Analysis of acetic acid-inducedwhitening of high-grade squamous intraepithelial lesions,” J. Biomed.Opt. , 6, 397–403.
[24]S. Y. Park, M. Follen, A. Milbourne, H. Rhodes, A. Malpica, N.Mackinnon, C. MacAulay, M. K. Markey, and R. Richards-Kortum(2008),“Automated image analysis of digital colposcopy for the detection ofcervical noplasia,” J. Biomed. Opt. 13, 014029-1–014029-10.
[25]P. J. Rousseeuw(1987), “Silhouettes: A graphical aid to the interpretation and validation of cluster analysis,” Computational and Applied Mathematics, 20, 53-65.
[26]R. Reid and P. Scalzi(1985), “Genital warts and cervical cancer. VII. Animproved colposcopic index for differentiating benign papillomaviralinfections from high-grade cervical intraepithelial neoplasia,” Am. J.Obstet. Gynecol. , 153, 611–618.
[27]J. W. Sellors, and R. Sankaranarayanan(2003), eds. Colposcopy and Treatment of Cervical Intraepithelial Neoplasia. A Beginner’s Manual. Lyon, France: International Agency for Research on Cancer.
[28]R. Sankaranarayanan, R. Rajkumar, R. Theresa, P. O. Esmy, C.Mahe, K. R. Bagyalakshmi, S. Thara, L. Frappart, E. Lucas, R. Muwonge,S. Shanthakumari, D. Jeevan, T. M. Subbarao, D. M. Parkin,and J. Cherian(2004), “Initial results from a randomized trial of cervicalvisual screening in rural south India,” Int. J. Cancer, 109, 461–467.
[29]A. K. Tain, M. N. Murthy, and P. J. Flynn(1999), Data clustering: a review, ACM Comput. Survey, 31, 3, 264–323.
[30]A. Ultsch(2007), “Emergence in Self-Organizing Feature Maps,” In Proceedings Workshop on Self-Organizing Maps (WSOM ''07). Bielefeld, Germany.
[31]S. Yang, J. Guo, P. King, Y. Sriraja, S. Mitra, B. Nutter, D. Ferris, M.Schiffman, J. Jeronimo, and R. Long(2004), “A multi-spectral digital cervigramanalyzer in the wavelet domain for early detection of cervicalcancer,” Proc. SPIE 5370, 1833–1844.