臺灣博碩士論文加值系統

子宮頸癌為台灣地區癌症的致命疾病之一，其發生率高居五大婦癌的榜首。傳統上以子宮頸抹片檢查為預防子宮頸癌最好的方式，一般配合子宮頸陰道鏡影像檢查，提供醫師在細胞型態、癌化程度與是否需要進一步施行組織切片檢查的主要依據。本研究主要為『子宮頸抹片異常細胞辨識之初期研究』，希望透過影像處理技術的應用，分析與子宮頸癌發生之可能相關細胞型態及特徵參數，藉以提供臨床醫師在子宮頸癌前期之診斷輔助。
本研究直接擷取子宮頸抹片影像作分析，將影像分為彩色影像與灰階影像兩部分進行。在彩色影像部分，透過RGB與HIS模型運算所得之參數提供不同損傷程度之細胞其表現差異。而在灰階影像部分，針對影像運用Histogram Equalization提高影像之對比度，利用灰度伴隨矩陣分析影像之紋理表現，最後由主動輪廓模型圈選出欲分析之細胞，計算其細胞核大小、核質比及細胞輪廓長度等參數。研究過程中，利用假體影像及標準圖譜共30張影像來測試系統之參數準確性與可行性，並針對臨床子宮頸癌病例共149張影像進行實際診斷鑑別及分析探討。此外，也藉由臨床醫師之協助比較系統與傳統人工鑑別細胞正常或異常的差異性。
初步結果顯示，系統不論在薄層抹片或傳統抹片影像上初步辨識細胞正常或是異常時，其Accuracy、Sensitivity與Specificity值均為1，表示系統在辨識正常細胞與異常細胞上具有良好的鑑別力。而在異常細胞鑑別型態部分，薄層抹片其Accuracy、Sensitivity與Specificity值分別為0.9、1與0.8；而傳統抹片其Accuracy、Sensitivity與Specificity值分別為0.841、0.905與0.783。導致Accuracy與Specificity不高的原因是在薄層抹片分析中有一筆FP值，因其細胞核質比較大，因此系統將該細胞判斷為HSIL；而傳統抹片中有2張HSIL的影像，其在特徵表現上與LSIL近似，因此系統鑑別錯誤；而將系統鑑識結果與人工比對發現，系統可以有效判斷肉眼較無法辨識之異常細胞。
整體而言，系統之完成能夠提供子宮頸癌細胞之形態分析與初步診斷，並能實際輔助臨床上之子宮頸癌細胞鑑別診斷，資料庫之建立亦可提供子宮頸癌治療前後之評估。未來可增加病例數與其他特徵參數之收集，使貝氏網路之訓練更加具有意義以提高系統之正確率，亦可加入異常細胞自動偵測來提高系統之效率。

Cervical cancer is the one of the deadly diseases of cancers in Taiwan, and its occurrence rate is the top of five women cancers. Traditionally, Pap smear is the best treatment for preventing cervical cancer. Originally cooperation with cervical colposcopy to provide doctors cell types, degree of cancer or whether to apply tissue section. This study is focused on the primarily study of Pap smear abnormal cells recognition. We hope to analyze cell types and characteristics parameters relation to cervical cancer via the allocations in image processing to provide diagnostic assistant for clinicians in cervical pre-cancer.
Pap smear images were caught to analyze in two different models. In color images, the parameters obtained via RGB and HIS model calculating provided the difference performance between different degree lesion cells. In grey images, histogram equalization was applied to images to enhance the contrast of images and co-occurrence matrix was used to analyze the textures of images. Finally by applied the active contour model to circle the interested cell, and then its cell nucleus’s size, N/C Ratio and cytoplasm’s path were calculated. To totally 30 phantom images and standard plates were used to train this system and 149 clinical cases were used to test the accuracy and feasibility of the system. Furthermore, through the help of clinicians, the comparison between the system and traditional method for the differences in distinguishing the normal and abnormal cells also done.
Preliminary results showed the accuracy, sensitivity and specificity were 1 in distinguishing normal and abnormal cells for ThinPrep and Pap smear images. It indicated the system has good identification in distinguishing normal and abnormal cells. In discriminating cells’ type, the accuracy, sensitivity and specificity of ThinPrep were 0.9, 1 and 0.8 individually which due to a FN case caused by its high N/C Ratio, and the accuracy, sensitivity and specificity of Pap smear were 0.841, 0.905 and 0.783 individually which due to two HSIL cases’ features are similar to LSIL. To compare our system with artificial detection can observe our system can diagnose abnormal cells identified by naked eyes.
In conclusion, the accomplishment of system can provide the cervical cancer cells’ type analysis and preliminary diagnosis, and practically assist the clinical cervical cancer cells’ discrimination diagnosis. The developed database also can provide the estimation of the treatment before and after. In the future, to increase cases and other characteristic parameters collection can make the Bayesian’s training more meaningful to improve the accuracy of the system, and also can add automatically detect for abnormal cells to raise the system’s efficiency.

中文摘要 -----------------------------------------------------------------------------------------------I
英文摘要 ---------------------------------------------------------------------------------------------III
誌 謝 --------------------------------------------------------------------------------------------------V
目 錄 ------------------------------------------------------------------------------------------------- VI
圖索引 ------------------------------------------------------------------------------------------------IX
表索引 -------------------------------------------------------------------------------------------------X

第一章 緒論 ------------------------------------------------------------------------------------------1
1-1 前言 ---------------------------------------------------------------------------------------------1
1-2 研究背景 ---------------------------------------------------------------------------------------2
1-3 文獻回顧 ---------------------------------------------------------------------------------------3
1-4 研究目的 ---------------------------------------------------------------------------------------7
1-5 論文架構 ---------------------------------------------------------------------------------------8

第二章 理論基礎 ------------------------------------------------------------------------------------9
2-1 子宮頸疾病 ------------------------------------------------------------------------------------9
2-1-1 發炎與炎症疾病 -------------------------------------------------------------------------10
2-1-1.1 細菌 ------------------------------------------------------------------------------------10
2-1-1.2 黴菌 ------------------------------------------------------------------------------------11
2-1-1.3 病毒 ------------------------------------------------------------------------------------11
2-1-2 浸潤前期與浸潤鱗狀細胞贅瘤 -------------------------------------------------------11
2-1-2.1 微浸潤鱗狀細胞癌 ------------------------------------------------------------------11
2-1-2.2 一般浸潤性鱗狀細胞癌 ------------------------------------------------------------12
2-1-2.3 乳突狀鱗狀細胞贅瘤 ---------------------------------------------------------------12
2-1-3 人類乳突病毒（HPV） ----------------------------------------------------------------12
2-1-3.1 HPV病毒構造 -----------------------------------------------------------------------13
2-1-3.2 HPV的感染 --------------------------------------------------------------------------13
2-1-3.3 HPV病毒感染與子宮頸鱗狀上皮癌前病變的特徵 --------------------------14
2-2 子宮頸抹片影像 -----------------------------------------------------------------------------15
2-2-1 子宮頸細胞抹片採樣 -------------------------------------------------------------------15
2-2-2 子宮頸細胞抹片種類 -------------------------------------------------------------------17
2-2-3 抹片的判讀 -------------------------------------------------------------------------------18
2-2-4 抹片診斷之準確性 ----------------------------------------------------------------------19
2-2-5 抹片偽陰性的原因 ----------------------------------------------------------------------20
2-3 抹片影像之細胞病理特徵與診斷標準 --------------------------------------------------21

2-3-1 鱗狀上皮病變 ----------------------------------------------------------------------------21
2-3-1.1 鱗狀上皮內病變（Squamous intraepithelial lesion） ------------------------21
2-3-1.2 鱗狀上皮癌（Squamous cell carcinoma） -------------------------------------22
2-3-2 腺體細胞病變 ----------------------------------------------------------------------------23
2-3-2.1 非典型子宮內頸細胞可能是腫瘤（probably neoplastic） ------------------24
2-3-2.2 子宮內頸腺癌（endocervical adenocarcinoma） ------------------------------24
2-4 子宮頸癌之臨床分期 -----------------------------------------------------------------------24
2-5 影像處理技術 --------------------------------------------------------------------------------26
2-5-1 影像分割 ----------------------------------------------------------------------------------26
2-5-1.1 直方圖處理（Histogram process） ----------------------------------------------26
2-5-1.2 直方圖均等化（Histogram equalization） -------------------------------------26
2-5-2 紋理分析 ----------------------------------------------------------------------------------27
2-5-2.1 高斯平滑 ------------------------------------------------------------------------------27
2-5-2.2 灰度伴隨矩陣 ------------------------------------------------------------------------27
2-5-3 主動輪廓模型（Active Contour Model） -----------------------------------------28
2-6 彩色影像特徵擷取原理 --------------------------------------------------------------------29
2-6-1 RGB彩色模型 ---------------------------------------------------------------------------30
2-6-2 HIS彩色模型 ----------------------------------------------------------------------------30
2-7 貝氏網路 --------------------------------------------------------------------------------------31
2-8 評估方式 --------------------------------------------------------------------------------------32

第三章 研究架構及方法 ------------------------------------------------------------------------34
3-1 研究材料與設備 -----------------------------------------------------------------------------34
3-1-1 研究材料 ----------------------------------------------------------------------------------34
3-1-2 研究設備 ----------------------------------------------------------------------------------35
3-2 研究流程 --------------------------------------------------------------------------------------35
3-3 研究步驟 --------------------------------------------------------------------------------------37
3-3-1 影像前級處理 ----------------------------------------------------------------------------37
3-3-2 主動輪廓模型 ----------------------------------------------------------------------------38
3-3-3 子宮頸抹片細胞特徵參數分析 -------------------------------------------------------39
3-3-4 資料庫與貝氏網路的建立 -------------------------------------------------------------40
3-3-5系統測試 -----------------------------------------------------------------------------------41
3-4 系統評估方式 --------------------------------------------------------------------------------43

第四章 結果與討論 ------------------------------------------------------------------------------46
4-1 系統正確性評估 -----------------------------------------------------------------------------46
4-1-1 特徵參數選取 ----------------------------------------------------------------------------46
4-1-1.1 不同損傷程度細胞在彩色特徵參數上之表現 ---------------------------------47
4-1-1.2不同損傷程度細胞在彩色特徵參數上之表現 ---------------------------------49
4-1-2 主動輪廓模型之正確性評估 ----------------------------------------------------------51
4-1-2.1 不同α值之圈選結果 ---------------------------------------------------------------51
4-1-2.2 不同β值之圈選結果 ---------------------------------------------------------------52
4-1-2.3 不同γ值之圈選結果 ---------------------------------------------------------------53
4-1-2.4 特徵參數計算結果評估 ------------------------------------------------------------53
4-1-3 主動輪廓模型圈選不同損傷程度之實際細胞影像結果 -------------------------54
4-2 影像資料選擇 --------------------------------------------------------------------------------58
4-3 貝氏網路訓練之建立 -----------------------------------------------------------------------59
4-4 異常細胞偵測分析之結果及病理探討 --------------------------------------------------61
4-4-1 細胞之色彩意義 -------------------------------------------------------------------------62
4-4-2 細胞之紋理意義 -------------------------------------------------------------------------63
4-5 細胞鑑別之結果 -----------------------------------------------------------------------------64
4-5-1 辨識正常細胞與異常細胞之結果評估 ----------------------------------------------65
4-5-2 鑑別異常細胞型態之結果評估 -------------------------------------------------------66
4-6 系統之限制 -----------------------------------------------------------------------------------69
4-7 系統之介面 -----------------------------------------------------------------------------------69

第五章 結論與未來展望 ------------------------------------------------------------------------76
5-1 結論 --------------------------------------------------------------------------------------------76
5-2 未來展望 --------------------------------------------------------------------------------------78

參考文獻 ---------------------------------------------------------------------------------------------80

附錄 ---------------------------------------------------------------------------------------------------84

作者自述 ---------------------------------------------------------------------------------------------85
圖 索 引
圖2-1子宮頸………………………………………………………………………...………10
圖2-2 HIS座標系統的雙圓錐模型 30
圖2-3貝氏網路架構圖 32
圖3-1研究所採用之子宮頸抹片細胞影像 35
圖3-2系統流程圖 36
圖3-3影像前級處理之變化 38
圖3-4主動輪廓模型計算流程圖 38
圖3-5 HSIL濃染細胞影像 40
圖3-6 Para-basal cell影像 40
圖3-7系統測試流程圖 42
圖3-8統計分析流程圖 43
圖3-9系統評估參數示意圖 45
圖4-1薄層抹片影像各特徵參數值在不同損傷細胞上之比較 48
圖4-2傳統抹片影像各參數值在不同損傷細胞上之比較 50
圖4-3不同α值對相同影像輪廓圈選之影響 52
圖4-4不同β值對相同影像輪廓圈選之影響 52
圖4-5不同γ值對相同影像輪廓圈選之影響 53
圖4-6特徵參數計算介面 54
圖4-7 Normal影像在不同參數下圈選所得之結果 55
圖4-8 LSIL影像在不同參數下圈選所得之結果 56
圖4-9 HSIL影像在不同參數下圈選所得之結果 57
圖4-10不同損傷細胞之主動輪廓模型參數 58
圖4-11細胞叢聚之抹片影像 59
圖4-12不同損傷程度影像分析所得之參數結果 63
圖4-13薄層抹片FP之細胞影像 67
圖4-14二個FN之細胞影像 72
圖4-15 HSIL抹片細胞影像 69
圖4-16 LSIL抹片細胞影像 69
圖4-17系統主介面 71
圖4-18病人病史資料介面 71
圖4-19病人影像介面 72
圖4-20 RGB模型運算影像介面 72
圖4-21運算結果顯示介面 73
圖4-22 Histogram Equalization介面 73
圖4-23細胞邊界圈選分析介面 74
圖4-24診斷介面 74
表 索 引

表2-1 鱗狀上皮病變名稱對照表 21
表2-2 子宮頸癌臨床分期（FIGO） 25
表2-3 評估參數示意圖 33
表3-1 抹片上細胞之細胞核平均大小 40
表3-2 低度SIL與高度SIL的細胞學變化 44
表4-1 p值與權重之關係……………………………………………………………………60
表4-2 薄層抹片之正常細胞與異常細胞辨識結果評估 61
表4-3 傳統抹片之正常細胞與異常細胞辨識結果評估 61
表4-4 薄層抹片之異常細胞型態鑑別結果評估 61
表4-5 傳統抹片之異常細胞型態鑑別結果評估 61
表4-6 薄層抹片之正常細胞與異常細胞辨識結果評估 65
表4-7 傳統抹片之正常細胞與異常細胞辨識結果評估 65
表4-8 正常細胞與異常細胞辨識在系統與病理師比較之評估 66
表4-9 薄層抹片之異常細胞型態鑑別結果評估 67
表4-10 傳統抹片之異常細胞型態鑑別結果評估 67
表4-11 異常細胞型態鑑別結果之評估（去除2張FN影像） 67
表4-12異常細胞型態鑑別在系統與病理師比較之評估 69

[1] 臺灣地區歷年癌症死亡數統計，行政院衛生署，http://www.doh.gov.tw/statistic/死因統計/st4_91_1.htm，Oct，2003。
[2] 嚴孟祿、謝長堯，子宮頸癌在台灣，行政院衛生署80年度委託研究計畫成果報告，1991，台北。
[3] H.M. Cole, Human papillomavirus DNA testing in the management of cervical neoplasia, JAMA, 1993, pp.2975-2981.
[4] 盧凡，子宮頸癌與人類乳突狀病毒簡介，健康天地電子報，第7期， http://www.health.nsysu.edu.tw/007-7.htm，1996。
[5] 黃思誠，子宮頸癌，台灣大學附設醫院，http://ntuh.mc.ntu.edu.tw/cgi-bin/bvuser.dll/ehospital/index.jsp，2002。
[6] V. Dadeshidze, L. Olsson, and R. A. Domanik, “Segmentation of nuclear images in automated cervical cancer screening”, SPIE. Society for Optical Engineering,Vol.2622, pp.723-727, 1995.
[7] R. Narayanswamy, J. P. Sharp, J. Heather, R. J. Stevart, and K. L. Johnson, “Morphological feature detection for cervical cancer screening”, SPIE. Society for Optical Engineering, Vol.2424, pp.265-275, 1995.
[8] J.P. Thiran and B. Macq, “Morphological feature extraction for the classification of digital images of cancerous tissues”, IEEE Transactions on Biomedical Engineering, Vol.43, pp. 1011-1020, 1996.
[9] S. W. K. Chang, K. S. Leung, and W. S. F. Wong, “Expert system for the detection of cervical cancer cells using knowledge-based image analysis”, Artificial Intelligence in Medicine, Vol.8, pp.67-90,1996.
[10] J. L. Alcazar and M. Jurado, “Trans-vaginal color Doppler for predicting pathological response to preoperative chemo-radiation in locally advanced cervical carcinoma: A Preliminary study”, Ultrasound in Medicine and Biology, Vol.25, No.7, pp.1941-1945, 1999.
[11] 曾志仁，子宮頸癌篩檢的改進，長庚紀念醫院，http://www.cgmh.org.tw/intr/intr5/c6700/CJTSENG2/education/b2.htm。
[12] U. Fayyad, D. Haussler, and P. Stolorz, ”The data warehouse and data mining”, Communication of The ACM, Vol.39, No.11, pp.49-57, November 1996.
[13] I. Kononenko, I.Bratko, and E.Roskar, “Expert system in automatic learning of medical diagnostic rules”, International School for the Synthesis of Expert's Kknowledge Workshop, Bled, Slovenia, August 1984.
[14] S. Hojker, I. Kononenko, A.Jauk, V.Filder, and M.Porenta, “Expert system's development in the management of thyroid diseases”, Proc. European Congress for Nuclear Medicine, Milano, Sep, 1988.
[15] K.A. Horn, P. Compton, L. LazarusL, and J.R Quinlan , “An expert system for the interpretation of thyroid assays in a clinical laboratory”, The Australian Computer Journal, Vol.17, No.1, pp.7-11, 1985.
[16] H. Ragavan, L. Rendell, M.Shaw, and Tessmer A. , “Lookahead feature construction for learning hard concepts”, Proc.10th Intern. Conf. on Machine Learning, pp.252-259, June 1993.
[17] E. Roskar, P. Abrams, I. Bratko and A.Varsek, “CUDS-An expert system for the diagnostics of lower urinary Tract Disorders”, Journal of Biomedical Measurements, Informatics and Control, Vol.1, No.4, pp.201-204, 1986.
[18] M. Kukar, I. Kononenko, and T. Silvester, “Machine learning in prognosis of the femoral neck fracture recovery”, Artificial Intelligence in Medicine, Vol.8, pp.431-451, 1996.
[19] K. E. Burnthornton and L. denbrand, “Myocardial-infarction-pinpoint the key indicators in the 12-lead ECG using data mining”, Computer and Biomedical Research, Vol.31, Iss.4, pp293-303, 1998.
[20] J. M. McDonald, S. Brossstte, and S.M. Moser, “Pathology information system data mining leads to knowledge discovery”, Pathology & Laboratory Medicine, Vol.122, Iss.5, pp 409-411, 1998.
[21] S. E. Brossette, A. P. Sprague, J. M. Hardin, K. B. Waites, W. T. Jones, and S. A. Moser, ”Association rules and data mining in-hospital infection-control and public-health surveillance “, Journal of the American Medical Informatics Association, Vol.5, Iss.4, pp.373-381, 1998.
[22] 阮正雄、傅堯喜與徐正光，子宮頸抹片與陰道鏡圖譜，嘉洲出版社，民國91年1月，台北。
[23] LG Koss, Diagnostic cytology and its histopathology bases, 4thed J.B Lippincott Company, Philadelphia, pp1-4, 1961.
[24] 吳香達，子宮頸癌，茂昌圖書公司，民國91年9月，台北。
[25] 林宸生，數位信號-影像與語音處理，全華出版社，台北，民國86年12月。
[26] E. L. Chen , P. C. Chung , C. L. Chen , H. M. Tsai , C. I. Chang，“An automatic diagnostic system for CT liver image classification”, IEEE Transactions on Biomedical Engineering, Vol.456 , June 1998 , p 783 –794.
[27] 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”，中興大學電機工程研究所碩士論文，民國九十二年。
[28] J. Park and J. M. Keller,” Snakes on the watershed”, IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 23, No. 10, October 2001.
[29] Chao Han, Thomas S. Hatsukami, Jenq-Neng Hwang, and Chun Yuan，” A fast minimal path active contour model”, IEEE Transactions on Image Processing, Vol. 10, No. 6, June 2001.
[30] 繆紹綱，數位影像處理 活用-Matlab，全華科技圖書股份有限公司，p8-2~8-28，民國八十八年。
[31] 吳國禎，“資料探索在醫學影像資料庫之應用”，中原大學醫學工程研究所碩士論文，民國九十年。
[32] 范振添，“醫療影像傳輸及資料探索之系統開發”，中原大學醫學工程研究所碩士論文，民國九十一年。
[33] Beth Dawson-Saunders and Robert G. Trapp, Basic and Clinical Biostatistics, Prentice-Hall International Inc., p233, 1994.