臺灣博碩士論文加值系統

子宮頸癌是臺灣婦女之首要好發癌症，而子宮頸抹片是一項簡單、便宜又極方便的篩檢方法，它已公認可有效的檢驗初癌前病變、細胞異生及各種癌症。近年來由於衛生署大力推廣子宮頸癌之細胞抹片篩檢，使得此癌症之死亡率下降。在台灣，目前的抹片篩檢率為20-30%，提昇篩檢率將增加篩檢量與負擔，但人為的篩檢與評估，終究有一些限制，例如主觀意識、疲勞、時效性等等，難以完全克服，所以設計出更快、更好、更有效的科學方法來改善細胞篩檢品質，已成為近年來國際間爭相研究的主題之一，目前臨床上已通過運用一些電腦輔助的自動化子宮頸癌細胞篩檢系統，例如PAPNET system (Neuromedical Systems, Inc., NY, USA) 以及AutoPap (NeoPath, Inc., USA)來作篩檢及評估的工作，顯示此項研究的可行性及應用價值極高。然而由於它們的價格昂貴、或判讀與操作過程複雜，使得其在國內之使用率其普遍性極低，幾乎沒有實驗室負擔得起其費用以及所耗之工時。我們的研究是借由通過衛生署審查合格的實驗室，收集經傳統固定及染色處理之後的抹片，透過數位相機及高解析度之顯微鏡， 取得國人的子宮頸抹片細胞之數位化記錄，再將各類細胞之數位化記錄並自行研發出判讀軟體系統。經過測試此系統的細胞辨識能力極佳，檢測異常細胞檢測的敏感度(sensitivity)為92%，精確度(specificity) 為90%，正確率(accuracy) 為 91%。此研究不但證實國人在醫學工程上的實力，已可自製研發出價廉物美、多功能且操作容易的細胞分析軟體系統。產生的子宮頸細胞數位化資料庫，可供建立國人子宮頸正常及不正常細胞的基本數值，以便進一步發展全自動的子宮頸抹片細胞篩檢儀器，以減低人工篩檢的負擔與診斷的錯誤率，並作為癌症細胞教學研究、診斷與照會等多項功能使用。除外，此細胞分析軟體也可作為人體其他細胞或其他生物細胞型態分析研究等等功能，對我國生物醫學科技的研究發展，提供一項強而有力的資訊化影像檔案分析工具。

In Taiwan, the mortality of uterine cervical carcinoma, the most common female cancer, has effectively been reduced by the cervical smear screening programs. To develop more effective and scientific methods or instruments for improving the quality of screening is an important issue in the world. Techniques of digital image processing and computer-aided analysis system are introduced in this field. PAPNET system (Neuromedical Systems, Inc., NY, USA) and AutoPap (NeoPath, Inc., USA) are representatives in the market. However, the high cost and inconvenience may limit the use of these commercial instruments in Taiwan. The fast changing and improvement of the synthetic industry of Taiwan is famous in these decades. However, very few studies focusing on the cytologic image medical engineering are done in Taiwan. This project aims at developing a low-cost automated analysis system using low-cost, easily available instruments in the usual laboratories, and a cellular image analytic program accomplished by the cooperation of local cytopathologists, medical image-processing scientists. In our study, the morphological features of uterine cervical cells were initially recorded through high-resolution microscopy, digital camera and computer-aided cellular image analysis system to construct a reasonable-sized database. The automated cellular analytic technologies for inspecting and identifying abnormal smears are developed based on the image processing and fuzzy artificial intelligence. The accuracy of the developed system is quite excellent (sensitivity 92%, specificity 90%, and accurate rate 91%). The proposed system is expected to set up an image data base for further development of an automated screening system to share the
exhausted duty and reduce the misdiagnosis of manual screening. In addition it may also be applied for other functions, including teaching, training, diagnostic consultation and proficiency test for cervical smears and other cellular investigation. It is a new powerful cellular analytic tool for Taiwanese medical and biological engineering investigation.

壹、摘要 頁數
一、中文摘要…………………………………………………………5-6
二、英文摘要…………………………………………………………7-8
貳、前言………………………………………………………………9-13
參、材料與方法
一、抹片的收集與製作………………………………………………14-17
二、子宮頸抹片的篩選與診斷………………………………………18-25
三、子宮頸抹片細胞影像的擷取與處理……………………………26
四、子宮頸抹片細胞影像的分析……………………………………27
五、子宮頸抹片細胞影像的判讀與辨識軟體的能力測試…………28-29
六、實驗設計之整體流程簡圖與簡寫符號一覽表…………………30-31
肆、結果與討論
一、研究的重要性與意義……………………………………………32-33
二、子宮頸抹片的製作與細胞的選擇………………………………34
三、系統設備之組裝與特色…………………………………………35-37
四、電腦細胞分析系統之設計理論…………………………………38-40
五、子宮頸細胞影像資料庫的建立、分析與執行困難……………41-49
六、電腦分析及辨識子宮頸細胞影像之能力………………………50-57
伍、結論與建議………………………………………………………58-59
陸、參考文獻…………………………………………………………60-64
柒、論文發表…………………………………………………………65
捌、附錄
一、各類正常與不正常子宮頸細胞數位化量測的原始數值………66-68
二、CIAS基礎細胞影像分析系統介紹………………………………69-80
玖、表目錄
表一：文獻中子宮頸抹片正常與不正常細胞之型態與測量……………42
表二：總擷取影像檔案、所佔記憶體大小以及可分析的細胞數…目…44
表三：數位影像分析出的子宮頸抹片細胞之各項測量值………………45
表四：子宮頸抹片細胞的文獻紀錄與數位影像分析測量值之比對
表四(1)：細胞之形狀、細胞大小、細胞面積、核面積之比對………46
表四(2)：核直徑、核質比之比對………………………………………47
表四(3)：染色質分佈(細胞核顆粒粗糙度)、染色質強度之比對……48
表五：數位影像分析出的子宮頸抹片正常細胞之各項辨識結果……57
壹拾、圖目錄
圖一、用採檢器在婦女子宮頸括取剝落細胞…………………………15
圖二、以線型自動染色機進行柏式染色法染色………………………16
圖三(1)、顯微高解析數位相機擷取影像模組之顯微鏡與數位相機…18
圖三(2)：顯微高解析數位相機擷取影像模組之電腦規格……………19
圖四、淺層鱗狀上皮細胞………………………………………………19
圖五、中層鱗狀上皮細胞………………………………………………20
圖六、側基底鱗狀上皮細胞……………………………………………20
圖七、成熟鱗狀上皮化生細胞…………………………………………21
圖八、內頸腺體細胞……………………………………………………21
圖九、子宮內膜腺體細胞………………………………………………22
圖十、子宮內膜間質細胞………………………………………………22
圖十一、低危險度鱗狀表皮內病灶……………………………………23
圖十二(1)、高危險度鱗狀表皮內病灶，中度異生…………………24
圖十二(2)、高危險度鱗狀表皮內病灶，重度異生…………………25
圖十三(1)、細胞核面積大小之數位影像常態分布圖………………51
圖十三(2)、細胞核濃度之數位影像常態分布圖……………………52
圖十三(3)、細胞核粗糙度之數位影像常態分布圖…………………53
圖十三(4)、細胞核直徑之數位影像常態分布圖……………………54
圖十三(5)、細胞面積之數位影像常態分布圖………………………55
圖十三(6)、核質比之數位影像常態分布圖…………………………56

陸、參考文獻
1. Cancer Registry Annual Report, Republic of China,. Executive, Yuan, ROC, Department of Health, 1998.
2. Paley PJ. Screening for the major malignancies affecting women: Current guidelines. Am J OBS GYN. 2001;184:1021-1030.
3. Health and National Insurance Annual Statistics Information Service, Republic of China, Executive, Yuan, ROC, Department of Health, 1993-2001.
4. Hartmann KE, Nanda K, Hall S, Myers E. Technologic Advances for Evaluation of Cervical Cytology: Is Newer Better? OBS & GYN Survey. 2001;56:765-774.
5. Brown D, Garber AM. Cost-effectiveness of 3 methods of enhance the sensitivity of Papnicolaou testing. JAMA 1999;281:347-53.
6. Schechter CB. Cost-effectiveness of rescreening conventionally prepared cervical smears by PAPNET testing. Acta Cytol 1996;40:1272-82.
7. Spitzer M, Cervical screening adjuncts: Recent advances. Am J Obstet Gynecol 1998;179:544-6.
8. Doornewaard H, Woudt JM, Strubbe P, van de Seijp H, van den Tweel JG. Evaluation of PAPNET-assisted cervical rescreening. Cytopathology 1997;8:313-21.
9. Fetterman B, Pawlick G, Koo H, Hartinger J, Gilbert C, Connell S. Determining the utility and effectiveness of the Neopath autopap300QC system used routinely. Acta Cytol 1999;43:13-22.
10. Denaro TJ, Herrriman JM, Shapira O. PAPNET testing system - Technical update. Acta Cytol 1997;41:65-73.
11. Ashfag R, Saliger F, Solares B, Thomas S, Liu G, Liang Y, Saboorian MH, Evaluation of the PAPNET system for rescreening triage of cervicovaginal smears. Acta Cytol 1997;41:1058-64.
12. Kok MR, Boon ME. Consequences of neural network technology for cervical screening. Increase in diagnostic consistency and positive scores. Cancer 1996:78:112-7.
13. Wilbur DC, Prey MU, Miller WM, Pawlick GF, Colgan TJ. The AutoPap system for primary screening in cervical cytology. Comparing the results of a prospective, intended-uses study with routine manual practice. Acta Cytol 1998;42:214-20.
14. Bibbo M, Hawthorne C, Zimmerman B. Dose use of the AutoPap assisted primary screener improve cytologic diagnosis? Acta Cytol 1999;43:23-6.
15. Marshall CJ, Towe L, Bentz JS. Improved quality-control detection of false-negative Pap smears using the AutoPap 300 QC system. Diagn Cytopathol 1999;20:170-4.
16. Patten SF, Lee JamesSJ, Wilbur DC, et at. The AutoPap 300 QC System Multicenter Clinical Trials for use in quality control recreening of cervical smears. Cancer (Cancer cytopathol) 1997;81:337-47.
17. Naib ZM: Cytopathology. 4th ed. Bosten New York: Little, Brown and Company, 1996;7-65.
18. Kurman RJ, Solomon D. The Bethesda system for reporting cervical-vaginal cytologic diagnoses: Definition, criteria, and explanatory notes for terminology and specimen adequacy. Springer-Verlag New York, Inc, 1994.
19. Sawaya GF, Washington AE. Cervical cancer screening: which techniques should be used and why? Clin Obset Gynecol 1999;42:922-938.
20. Coleman DV. Evaluation of automated systems for the analysis of cervical smears. Cytopathol 1998;9:359-68.
21. Taylor LN, Gagnon M, Lange J, Lee T, Draut R, Kujawski E. Cytoview a prototype computer image-based Papanicolaou smear proficiency test. Acta Cytol 1999;43:1045-51.
22. Otsu N. A threshold selection method from gray-level histograms. IEEE Trans System. Man Cybern 1979; SMC-9:115-120.
23. Gonzalez RC, Wintz P. Digital Image Processing. Addisopm-Wesley, 1989.
24. M. Dai, Baylou P, Humbert L, Najim M. Image segmentation by a dynamic thresholding using edge detection based on cascaded uniform filters. Signal Process1996;52: 49-63.
25. Chen YS, Hwang HY, Chen BT. Color image analysis using fuzzy set theory. Proc Int Conf Image Processing 1995;1:242-245.
26. Kao CI, Kuo YH. A neural network model based on fuzzy classification concept. Proc Int Conf Neural Network 1992;2:727-732.
27. Xie XL, Beni G. A validity measure for fuzzy clustering. IEEE Trans Patt Anal & Mach Intell 1991;13:841-847.
28. Maclin S, Dempsey J. A neural network to diagnose liver cancer. Proc. IEEE Int. Conf. Neural Networks 1993;3:1492-1497.
29. Adali T, Wang Y. Probabilistic neural networks for medical image quantification. Proc. 1st Int. Conf. Image Processing 1994;3:889-892.
30. Kao CI, Kuo YH, A neural network model based on fuzzy classification concept. Proc. Int. Conf. Neural Networks 1992;2:727-732.
31. Haykin S. Neural network: a comprehensive foundation, Upper Saddle River, New Jersey : Prentice Hall, 1994.
32. Gonzalez RC, Woods RE, Digital Image Processing, Addison-Wesley, 1992.
33. Liu J, Tang YY, Cao YC. An evolutionary autonomous agents approach to image feature extraction. IEEE Trans Evolutionary Computation 1997;1: 2: 141-158.
34. Klir GJ, Yuan B. Fuzzy sets and fuzzy logic: Theory and application. NJ: Prentice Hall, 1995.
35. Krishnapuram R, Keller JM, Ma Y. Quantitative analysis of properties and spatial relations of fuzzy image regions. IEEE Trans Fuzzy Syst,
柒、論文發表
1. PC Huang, YT Chen, GH Hsiue. “The computer-aided cellular image analysis system for cervicovaginal smear”, World Congress for Chinese Biomedical Engineers, pp.118,2002.
2. PC Huang, YTChen, GH Hsiue. “The computer-aided cellular image analysis system for cervicovaginal smear”, The 4th Asian-Pacific Organization for Cell Biology Congress (APOCB), pp 105, 2002.