過去的肺腫瘤電腦輔助診斷(Computer-aided Diagnostic, CAD)系統主要可分為型態特徵與血流灌注特徵2種方法。本研究透過自動檢測及電腦視覺技術的使用，發展一套結合型態特徵與血流灌注特徵的肺腫瘤CAD系統。此一CAD系統具有2個顯著的特點，首先我們的系統包含一個有效的3D肺腫瘤半自動切割方法；其次是經過CAD實驗發現，僅需要使用注入顯影劑前及注入顯影劑後第90秒兩個時間點的CT掃瞄影像即可計算有效的血流灌注特徵。相較於傳統動態CT研究需要3~10次的CT掃瞄，本研究的方法縮短了CT掃瞄的處理時間，並且減少病患所需接受到的X光輻射劑量。除此之外，肺結核是台灣常見的良性腫瘤，而且過去的研究顯示，在臨床上肺結核容易被誤診為肺癌。然而本研究的CAD方法，將肺結核正確診斷為良性腫瘤的機率可達到92.9%，應可有效減少臨床誤診的可能性。
　　相較於使用3D影像的CAD系統，傳統使用2D影像的系統具有操作簡便的優點，而3D系統的優點是具有完整的肺腫瘤資訊。為了分析2D與3D系統的診斷效能差異，本研究亦進行一系列的實驗比較。同時我們亦建立一個結合2D與3D CAD方法的二階段分類器，以結合兩者的優點。實驗結果顯示擁有較多腫瘤資訊的3D系統，其診斷效果確實優於2D系統。而且透過本研究建立的二階段方法，將可在僅些微降低診斷正確率的情況下，有效減少需使用3D CAD系統進行診斷的肺腫瘤比例，以減少肺腫瘤診斷工作的複雜度。

Several computer-aided diagnostic (CAD) methods for solitary pulmonary nodules (SPNs) have been proposed, which can be divided into two major categories: (1) the morphometric CT method, and (2) the perfusion CT method. The first goal of this work is to introduce a neural network-based CAD method of lung nodule diagnosis by combining morphometry and perfusion characteristics by perfusion CT. The proposed approach has the following distinctive features. Firstly, this work develops a very efficient semi-automatic procedure to segment entire nodules. Secondly, reliable nodule classification can be achieved by using only two time-point perfusion CT feature measures (precontrast and 90 s). This greatly reduces the amount of radiation exposure to patients and the data processing time. As demonstrated in previous work, classification tuberculomas from malignancies has been considered to be a challenging task. However the diagnosis accuracy for tuberculomas reaches 92.9% by applying the proposed CAD method.
Another goal of this work is, by investigating the relative merits of 2D and 3D methods, to develop a two-stage approach that combines the simplicity of 2D and the accuracy of 3D methods. Experimental results show statistically significant differences between the diagnostic accuracy of 2D and 3D methods. The results also show that with a very minor drop in diagnostic performance the two-stage approach can significantly reduce the number of nodules needed to be processed by the 3D method and thus alleviates the computational demand.

論文目錄
目錄 I
圖目錄 IV
表目錄 VII
摘要 X
Abstract XI
第一章 緒論 1
1.1 前言 1
1.2 肺癌的臨床症狀 3
1.3 肺腫瘤 7
1.4 電腦斷層攝影(CT)簡介 8
1.5 文獻回顧 10
第二章 研究動機及目的 13
2.1 研究動機 13
2.2 研究目的 15
第三章 肺腫瘤診斷系統架構及方法 16
3.1 資料收集 17
3.2 腫瘤切割法 21
3.3 腫瘤特徵萃取 25
3.3.1 型態特徵 26
3.3.2 血流灌注特徵 33
3.4 設計分類器 36
3.4.1. 類神經網路 37
3.4.2. 委員會機器 38
3.5 實驗方法設計 42
3.5.1. 直接學習法(Direct Learning Method, DLM) 42
3.5.2. 單點移出法(Leave-one-out Strategy, LOOS) 42
3.5.3. 隨機取樣法(Random Sampling Method, RSM) 43
第四章 3D CAD系統實驗結果 45
4.1 型態特徵測試 45
4.2 血流灌注特徵測試 46
4.3 血流灌注及型態特徵的整合測試 48
4.4 肺結核分類工作分析 49
4.5 腫瘤尺寸與鑑別率分析 50
4.6 與前人方法的比較 53
第五章 2D與3D CAD系統的效能分析 56
5.1 實驗資料 57
5.2 2D CAD系統的實驗方法 57
5.3 2D系統與3D系統之良、惡性診斷能力比較 58
5.4 二階段決策方法 – 2D與3D CAD系統之整合 60
5.4.1 兩階段決策方法的概念及簡介 61
5.4.2 時間效率提升指標(Time Efficiency Index, TEI) 65
5.4.3 實驗結果 66
第六章 結論 69
參考文獻 72
附錄I 78
附錄II 81
附錄III 86

參考文獻
Armato, S.G. 3rd and MacMahon, H., “Automated lung segmentation and computer-aided diagnosis for thoracic CT scans,” Int Congr Ser, vol. 1259, pp. 977-982, 2003.
Awai, K., Murao, K., Ozawa, A., Nakayama, Y., Nakaura, T., Liu, D., Kawanaka, K., Funama, Y., Morishita, S., and Yamashita, Y., “Pulmonary nodules: estimation of malignancy at thin-section helical CT-effect of computer-aided diagnosis on performance of radiologists,” Radiol., vol. 239, no. 1, pp. 276-284, Apr. 2006.
Breiman, L., “Bagging predictors,” Mach. Learn., vol. 24, no. 2, pp. 123-140, 1996.
Chang, Y.C., Cheng, H.H., Yeh, C.F., Lee, W.J., Yang, P.C., Liu, M.H., and Su, J.L., “Textural analysis of lung nodule smaller than 3 cm with dynamic contrast enhancement,” in Proc. 1st World Congress of Thoracic Imaging and Diagnosis in Chest Disease, Italy, p. 265, 2005.
Chawalparit, O., Charoensak, A., and Chierakul, N., “HRCT of pulmonary tuberculosis mimics malignancy: a preliminary report,” J. Med. Assoc. Thai., vol. 89, no. 2, pp. 190-195, Feb. 2006.
Dietterich, T.G., “An experimental comparison of three methods for constructing ensembles of decision trees: bagging, boosting, and randomization,” Mach. Learn., vol. 40, no. 2, pp. 139-157, 2000.
Downey, T.J., Meyer, D.J., Price, R.K., and Spitznagel, E.L., “Using the Receiver Operating Characteristic to Asses the Performance of Neural Classifiers,” Int. Joint Conf. Neural Networks, vol. 5, pp. 3642-3646, 1999.
Efron, B. and Tibshirani, R., “An Introduction to Bootstrap,” Chapman & Hall, London, 1993.
Folkman, J., Merler, E., Abernathy, C., and Williams, G. “Isolation of a tumor factor responsible for angiogenesis,” J. Exp. Med., vol. 133, no. 2, pp. 275-288, Feb. 1971.
Feng, S.T., Chen, J.D., Meng, Q.F., Yang, X.F., Xie, H.B., and Yan, C.G., “Imaging features of lung carcinoma, pulmonary tuberculoma, and inflammatory pseudotumor on helical incremental dynamic CT scan--a report of 44 cases,” Ai Zheng, vol. 25, no. 3, pp. 348-351, 2006.
Fuchs, T.O.J., Kachelriess, M., and Kalender, W.A., “Fast Volume Scanning Approaches by X-Ray-Computed Tomography,” In: Proc IEEE Comput Soc Bioinform Conf, vol. 91, no. 10, pp.1492-502, 2003.
Kanazawa, K., Kawata, Y., Niki, N., Satoh, H., Ohmatsu, H., and Kakinuma, R., “Computer-aided diagnosis for pulmonary nodules based on helical CT images,” Comput Med Imaging Graph, vol. 22, no. 2, pp. 157-67, 1998.
Gaeta, M., Volta, S., Bartiromo, G., Stroscio, S., Minutoli, F., and Barone, M., "Contrast-enhanced study of solitary pulmonary nodules with thin-section computed tomography," Radiol. Med. (Torino), vol. 94, no. 3, pp. 189-192, Sept. 1997.
Haykin, S., “Neural network: a comprehensive foundation, 2nd Edition,” Prentice Hall, 1999.
Henschke, C.I. and Yankelevitz, D.F., “CT Screening for Lung Cancer,” Radiologic Clincs of North America, vol. 38, pp. 487-495, 2000.
Hu, S., Hoffman, E.A., and Reinhardt, J.M., “Automatic lung segmentation for accurate quantitation of volumetric x ray CT images,” IEEE Trans Med Imag, vol. 20, no. 6, pp. 490-8, 2001.
Jeong, Y.J., Lee, K.S., Jeong, S.Y., Chung, M.J., Shim, S.S., Kim, H., Kwon, O. J., and Kim, S., "Solitary pulmonary nodule: characterization with combined wash-in and washout features at dynamic multi-detector row CT," Radiol., vol. 237, no. 2, pp. 675-683, Nov. 2005.
Kan, C. and Srinath, M.D., “Invariant character recognition with Zernike and orthogonal Fourier-Mellin moments,” Pattern Recognition, vol. 35, pp. 143-154, 2002.
Kawata, Y., Niki, N., Ohmatsu, H., and Moriyama, N., “Example-Based Assisting Approach for Pulmonary Nodule Classification in Three-Dimensional Thoracic Computed Tomography Images,” Acad. Radiol., vol. 10, no. 12, pp. 1402-15, 2003.
Kuhnigk, J.M., Dicken, V., Bornemann, L., Bakai, A., Wormanns, D., and Krass S., “Morphological Segmentation and Partial Volume Analysis for Volumetry of Solid Pulmonary Lesions in Thoracic CT Scans,” IEEE Trans Med Imag, vol. 25, no. 5, pp. 417-34, 2006.
Li, F., Sone, S., Abe, H., MacMahon, H., and Doi, K., “Malignant versus benign nodules at CT screening for lung cancer: comparison of thin-section CT findings,” Radiol., vol. 233, no. 3, pp. 793-798, 2004.
Lo, C.H. and Don, H.S. “3-D moment forms: their construction and application to object identification and positioning,” IEEE Trans. Pattern Anal. and Machine Intell., vol. 11, no. 10, pp. 1053-1064, 1989.
McNitt-Gray, M.F., Wyckoff, N., Sayre, J.W., Goldin, J.G., and Aberle, D.R., “The effects of co-occurrence matrix based texture parameters on the classification of solitary pulmonary nodules imaged on computed tomography,” Comput Med Imaging Graph, vol. 23, no. 6, pp. 339-48, 1999.
Miles, K. A., “Tumour angiogenesis and its relation to contrast enhancement on computed tomography: a review,” Eur. J. Radiol., vol. 30, no. 3, pp. 198-205, 1999.
Miles, K. A., “Perfusion CT for the assessment of tumor vascularity: which protocol?,” Br. J. Radiol., vol. 76, no. 1, pp. S36-S42, 2003.
Miles, K.A., Charnsangavej, C., Lee, F.T., Fishman, E.K., Horton, K., Lee, T.-Y., “Application of CT in the investigation of angiogenesis in oncology,” Acad. Radiol., vol. 7, no. 10, pp. 840-850, 2000.
Mulshine, J.L., Hong, S., Martinez, A., Tauler, J., Avis, I., Tockman, M.S., De Luca, L.M., Placke, M.E., and Cuttitta, F., “Moving to the Routine Management of Pre Symptomatic Lung Cancer,” Lung Cancer, vol. 34, suppl. 2, pp. S1-S5, 2001.
Mori, K., Niki, N., Kondo, T., Kamiyama, Y., Kodama, T., Kawada, Y., and Moriyama, N., "Development of a novel computer-aided diagnosis system for automatic discrimination of malignant from benign solitary pulmonary nodules on thin-section dynamic computed tomography," J. Comput. Assist. Tomogr., vol. 29, no. 2, pp. 215-222, Mar. 2005.
Ng, B., Abugharbieh, R., Huang, X., and McKeown, M.J., “Characterizing fMRI activations within regions of interest (ROIs) using 3D moment invariants,” In: Proc IEEE 2006 Conf Comput Vision Patt Recog(CVPR), 2006.
Novotni, M. and Klein, R. “Shape retrieval using 3D Zernike descriptors,” Computer-Aided Design, vol. 36, no. 11, pp. 1047-1062, 2004.
Opitz, D. and Maclin, R. “An empirical evaluation of bagging and boosting for artificial neural networks,” in Conf. Rec. 1997 IEEE Int. Conf. Neural Network, pp. 546-551, 1997.
Parikh, R., Mathai, A., Parikh, S., Sekhar, G C., and Thomas, R., “Understanding and using sensitivity, specificity and predictive values,” Indian J. Ophthalmol, vol. 56, no. 1, pp. 45-50, 2008.
Patel, J. D., “Lung Cancer in Women,” J. Clin. Oncol., vol. 23, no. 14, pp. 3212-3218, May, 2005.
Petkovska, I., Shah, S.K., McNitt-Gray, M.F., Goldin, J.G., Brown, M.S., Kim, H.J., Brown, K., and Aberle, D.R., “Pulmonary nodule characterization: a comparison of conventional with quantitative and visual semi-quantitative analyses using contrast enhancement maps,” Eur. J. Radiol., vol. 59, no. 2, pp. 244-252, Apr. 2006.
Petty, T.L., “The early diagnosis of lung cancer,” Disease-a-month : DM., vol. 47, pp. 204-264, 2001.
Reeves, A.P., Chan, A.B., Yankelevitz, D.F., Henschke, C.I., Kressler, B., and Kostis, W.J., “On Measuring the Change in Size of Pulmonary Nodules,” IEEE Trans Med Imag, vol. 25, no. 4, pp. 435-50, 2006.
Sadjadi, F.A. and Hall, E.L., “Three-dimensional moment invariants,” IEEE Trans. Pattern Anal. and Machine Intell., vol. 2, no. 2, pp. 127-136, 1980.
Shah, S.K., McNitt-Gray, M.F., De Zoysa, K.R., Sayre, J.W., Kim, H.J., Batra, P., Behrashi, A., Brown, K., Greaser, L.E., Park, J.M., Roback, D.K., Wu, C., Zaragoza, E., Goldin, J.G., Suh, R.D., Brown, M.S., and Aberle, D.R., “Solitary pulmonary nodule diagnosis on CT: results of an observer study,” Acad. Radiol., vol. 12, no. 4, pp. 496-501, Apr. 2005.
Shah, S.K., McNitt-Gray, M.F., Rogers, S.R., Goldin, J.G., Suh, R.D., Sayre, J.W., Petkovska, I., Kim, H.J., and Aberle, D.R., “Computer aided characterization of the solitary pulmonary nodule using volumetric and contrast enhancement features,” Acad. Radiol., vol. 12, no. 10, pp. 1310-1319, Oct. 2005.
Siegelman, S.S., Khouri, N.F., Leo, F.P., Fishman, E.K., Braveman, R.M., and Zerhouri, E.A., “Solitary pulmonary nodules: CT assessment,” Thoracic Radiology, vol. 160, pp. 307-312, 1986.
Suzuki, K., Li, F., Sone, S., and Doi, K., “Computer-aided diagnostic scheme for distinction between benign and malignant nodules in thoracic low-dose CT by use of massive training artificial neural network,” IEEE Trans. Med. Imag., vol. 24, no. 9, pp. 1138-1150, Sep. 2005.
Swensen, S.J., Viggiano, R.W., Midthun, D.E., Muller, N.L., Sherrick, A., Yamashita, K., Naidich, D.P., Patz, E.F., Hartman, T.E., Muhm, J.R., and Weaver, A.L. "Lung nodule enhancement at CT: multicenter study," Radiol., vol. 214, no. 1, pp. 73-80, Jan. 2000.
Tan, B.B., Flaherty, K.R., Kazerooni, E.A., and Iannettoni, M.D. "The solitary pulmonary nodule," Chest, vol. 123, no. 1 suppl, pp. 89S-96S, Jan. 2003.
Tateishi, U., Kusumoto, M., Akiyama, Y., Kishi, F., Nishimura, M., and Moriyama, N., "Role of contrast-enhanced dynamic CT in the diagnosis of active tuberculoma," Chest, vol. 122, pp. 1280-1284, Oct. 2002.
Viggiano, R.W., Swensen, S.J., and Rosenow, E.C., “Evaluation and Management of Solitary and Multiple Pulmonary Nodules,” Clinics in the Chest Medicine, vol. 13, pp. 83-95, 1992.
Witschi, H., “A Short History of Lung Cancer,” Toxicological Sciences, vol. 64, pp. 4-6, 2001.
Yi, A., Lee, K.S., Kim, E.A., Han, J., Kim, H., Kwon, O.J., Jeong, Y.J., and Kim, S., “Solitary pulmonary nodules: dynamic enhanced multi-detector row CT study and comparison with vascular endothelial growth factor and microvessel density,” Radiol., vol. 233, no. 1, pp. 191-199, Oct. 2004.
Zwirewich, C.V., Vedal, S., Miller, R.R., and Müller, N.L., “Solitary Pulmonary Nodule: High-resolution CT and Radiologic-pathologic Correlation,” Thoracic Radiology, vol. 179, pp. 469-476, 1991.
行政院衛生署，「民國九十四年癌症登記報告」，http://crs.cph.ntu.edu.tw/crs_c/annual.html，2007。
行政院衛生署，「民國九十二年癌症登記報告」，http://crs.cph.ntu.edu.tw/crs_c/annual.html，2006。
國家衛生研究，「肺癌診治共識」，http://sars.nhri.org.tw/publish/lungcancer.php，1998。
張基晟，「肺癌的診斷方法和使用儀器」，http://www3.vghtc.gov.tw:8082/cm/Asthma/old/Cancer/simenl02.htm，2000。
楊泮池，「女性國人健康的頭號殺手—肺腺癌」，http://www.nhri.org.tw/nhri_org/pr/newspage/20040721_3.htm，2004。