臺灣博碩士論文加值系統

多發性硬化症(Multiple Sclerosis, MS)是一種發生於中樞神經系統(包括腦部以及脊髓)的疾病，因為可能同時有多處的神經出現髓鞘質脫失受損的情形，所以稱為「多發性」，患者可能會出現行動不便，視力受損，疼痛等症狀。神經科醫生通常利用核磁共振造影(MRI)技術來診斷此病症，透過MR影像，醫生可以辨識出多發性硬化症病灶發生之位置，但其病灶在影像上通常僅占少數像素，在影像辨識上不易以人工判斷，因此本研究配合影像中的像素強度、位置、與鄰域資訊，使用統計模型來找尋MR影像上多發性硬化症病灶之位置，包括條件隨機域、馬可夫隨機域、簡單貝氏分類器，同時整合3種方法，期望更精確尋找多發性硬化症，並且去驗證過去研究中所使用的3種影像T1、T2和FLAIR之間的相互搭配。實驗結果發現，影像以T2和FLAIR的搭配績效最好，而整合3種方法上則能更精準的尋找出多發性硬化症病灶，顯示本研究提出的統計模型能夠有效辨別多發性硬化症病灶，可以降低醫生在影像辨識上所花費的時間。

Multiple sclerosis is a disease in the central nervous system including the brain and spinal cord. Patients' nerve has many demyelination damaged locations, so called multiple. Patients will appear mobility, impaired vision, pain and other symptoms. Neurologists often use magnetic resonance imaging (MRI) technology to diagnose this disease. Doctors can identify the location of the occurrence of multiple sclerosis lesions by MRI. There are usually only a small number of pixels on the image belonging to MS lesions, so the image recognition is not easy for human judgment. In this study, we applied statistical models with pixel intensity, location, and neighborhood information to find the location of MS on MRI. The methods we used are conditional random fields, Markov random fields, Naïve Bayes classifier. We integrated these three methods to increase precise rate in finding MS lesions. Three kinds images including T1, T2 and FLAIR are cross combined for performance evaluation. Experiment results show that combining T2 and FLAIR images has the best performance. The proposed method has more accurate to find out the MS lesions. The results show the proposed method can identify MS lesions effectively and can reduce time consumed in image recognition for doctors.

摘要 I
Abstract II
致謝 III
目錄 IV
表目錄 VI
圖目錄 VII
第一章 、緒論 1
第一節 、研究背景 1
第二節 、研究動機 5
第三節 、研究目的 5
第四節 、研究架構 6
第二章 、文獻探討 9
第一節 、腦部疾病與多發性硬化症病灶 9
第二節 、特徵 11
一、 強度 12
二、 位置 13
三、 鄰域 13
第三節 、統計模型 14
一、 簡單貝氏分類器(NBC) 14
二、 馬可夫隨機域(MRF) 15
三、 條件隨機域(CRF) 16
第三章 、研究方法 17
第一節 、研究方法流程圖 17
第二節 、SPM分割 19
第三節 、特徵處理 20
一、 強度 20
二、 位置 21
三、 鄰域 23
第四節 、簡單貝氏分類器 24
第五節 、馬可夫隨機域 26
第六節 、條件隨機域 27
第七節 、整合 31
第八節 、評估方法 31
第四章 、研究結果 33
第一節 、程式開發環境 33
第二節 、實驗資料來源 33
第三節 、參數設定 34
一、 簡單貝氏分類器(NBC) 34
二、 馬可夫隨機域(MRF) 35
三、 條件隨機域(CRF) 36
第四節 、實驗結果與分析 36
一、 簡單貝氏分類器結果 36
二、 馬可夫隨機域結果 38
三、 條件隨機域結果 40
四、 本研究整合結果 42
五、 T1、T2、FLAIR 44
六、 T2、FLAIR 46
七、 T2 48
八、 統計模型比較 50
九、 影像組合比較 50
第五章 、結論 53
第一節 、結論與貢獻 53
第二節 、研究限制與未來研究方向 54
參考文獻 55

[1] 吳維彥，2006，『應用不定長度特徵之條件隨機域於口語不流暢語流修正模型』，台灣碩博士論文知識加值系統。
[2] 連子建，2006，『結合混合型離散化和挑選式簡易貝氏特徵選取來改善簡易貝氏分類器正確率之方法』，台灣碩博士論文知識加值系統。
[3] 蘇建源，2007，『條件式隨機域之圖形處理於人類動作辨識之應用』台灣碩博士論文知識加值系統。
[4] 張仁明，林達德，1999，『類神經網路與貝氏分類法應用於影像分割之比較研究』，農業機械學刊，第8卷，第3期，61-74頁。
[5] 小小神經學網http://www.dls.ym.edu.tw
[6] Gonzalez, R. C. , and Woods, R. E., “Digital Image Processing(2 ed),” Prentice-Hall Inc., Englewood Cliffs, NJ, 2002.
[7] Akselrod-Ballin, A., Galun, M., Gomori, J. M., Valsasina, P., Basri, R., and Brandt, A., "Automatic Segmentation and Classification of Multiple Sclerosis in Multichannel MRI," IEEE Transactions on Biomedical Engineering, Vol. 56, 2009, pp. 2641-2469.
[8] Karimaghaloo, Z., Shah, M., Francis, S. J., Arnold, D. L., Collins, D. L., and Arbel, T., "Automatic Detection of Gadolinium-Enhancing Multiple Sclerosis Lesions in Brain MRI Using Conditional Random Fields," IEEE Transactions on Medical Imaging, Vol. 31, No. 6, June 2012, pp.1181-1194.
[9] Ge, Y., "Multiple Sclerosis: The Role of MR Imaging," AJNR Am J Neuroradiol, 2006, pp. 1165–1176.
[10] Karimaghaloo, Z., Shah, M., Francis, S. J., Arnold, D. L., Collins, D. L., and Arbel, T., "Detection of gad-enhancing lesions in multiple sclerosis Using Conditional Random Fields," Springer-Verlag Berlin Heidelberg, 2010, pp. 41-48.
[11] Khayati, R., Vafadust, M., Towhidkhah, F., and Nabavi, S. M., "Fully automatic segmentation of multiple sclerosis lesions in brain MR FLAIR images using adaptive mixtures method and markov random field model," Computers in Biology and Medicine, Vol. 38, 2008, pp. 379-390.
[12] García-Lorenzo, D., Prima, S., Morrissey, S., and Barillot, C., "A robust expectation-maximization algorithm for multiple sclerosis lesion segmentation," MICCAI Workshop: 3D Segmentation in the Clinic: A Grand Challenge II, MS lesion segmentation, 2008, pp. 1-9
[13] Shiee, N., Bazin, P., and Pham, D.L., "Multiple sclerosis lesion segmentation using statistical and topological atlases," in: Grand Challenge Workshop: multiple sclerosis lesion segmentation Challenge, 2008, pp. 1-10.
[14] Bricq, S., Collet, C., Armspach, J. P., "Unifying framework for multimodal brain MRI segmentation based on Hidden Markov Chains," Medical Image Analysis, Vol. 12, 2008, pp. 639-652.
[15] Kroon, D., Oort, E., and Slump, K., "Multiple sclerosis detection in multispectral magnetic resonance images with principal components analysis," in: Grand Challenge Workshop: multiple sclerosis lesion segmentation Challenge, 2008, pp. 1-14.
[16] Datta, S., Sajja, B.R., He, R., Gupta, R. K., Wolinsky, J. S., and Narayana, P. A., "Segmentation of gadolinium-enhanced lesions on MRI in multiple sclerosis," Journal of Magnetic Resonance Imaging, Vol. 25, 2007, pp. 932-937.
[17] Saha, S., and Bandyopadhyay, S., "A new point symmetry based fuzzy genetic clustering technique for automatic evolution of clusters," Information Sciences, 2009, pp. 3230–3246.
[18] Sajja, B. R., Datta, S., He, R., Mehta, M., Gupta, R. K., Wolinsky, J. S., Narayana, P. A., "Unified approach for multiple sclerosis lesion segmentation on brain MRI," Annals of Biomedical Engineering. Vol. 34, 2006, pp. 142–151.
[19] Goldberg-Zimring, D., Achiron, A. Miron, S., Faibel, M., and Azhari, H., "Automated detection and characterization of multiple sclerosis lesions in bra in MR images," Magnetic Resonance Imaging, Vol. 16, No. 3, 1998, pp. 311–318.
[20] Wu, Y., Warfield, S. K., Tan, I. L., Wells, W., Meier, D., Schijndel, R., Barkhof, F., and Guttmann, C., "Automated segmentation of multiple sclerosis lesion subtypes with multichannel MRI," NeuroImage Vol. 32, 2006, pp. 1205–1215.
[21] Khayati P., Vafadust, M., Towhidkhah, F., and Nabavi, S. M., "A novel method for automatic determination of different stages of multiple sclerosis lesions in brain MR FLAIR images," Computerized Medical Imaging and Graphics, Vol. 32, 2008, pp. 124-133.
[22] Anbeek, P., Vincken, K. L., Osch, M., Bisschops, R., and Grond, J., "Probabilistic segmentat ion of white matter lesions in MR imaging," NeuroImage Vol. 21, 2004, pp. 1037–1044.
[23] Subbanna, N., Shah, M., Francis, S. J., Narayannan, S., Collins, D. L., Arnold, D. L., and Arbel, T., "MS lesion segmentation using Markov Random Fields," in: Workshop Medical Image Analysis Multiple sclerosis, 2009, pp. 15–26.
[24] Han, Jiawei, Kamber, Micheline, and Pei, Jian, Data Mining: Concepts and Techniques(2 ed), Morgan Kaufmann, 2005.
[25] Leemput, K. V., Maes, F., Vandermeulen D., Colchester, A., and Suetens, P., "Automated Segmentation of Multiple Sclerosis Lesions by Model Outlier Detection" IEEE Transactions on Medical Imaging, Vol. 20, No. 8, 2012, pp.677-688.