臺灣博碩士論文加值系統

近年來，隨著醫療技術進步，腦部疾病不再是令人絕望的末日，研究顯示，隨著疾病不同，受到影響的腦部灰質與白質亦變化多端，就醫學影像檢查角度而言，大腦灰、白質組織性質相似，不易區分，因此如何準確分割出病變組織，無庸置疑是當前治療腦部病變的重要課題。
幸運的是，現今非侵入式檢查方法數以千計，不必再挨刀受苦檢查，而其中最重要被廣泛使用的便是核磁共振造影(Magnetic Resonance Imaging, MRI)，本篇論文便是針對MRI影像去進行分門別類，找出病灶所在。
影像分類需要一系列訓練點去區分，分為非監督式或監督式兩種方法，差別在於訓練點產生方式是未知或者預先得知資料，本篇論文選擇以非監督法演算法單形體體積生長演算法(simplex growing algorithm, SGA)找尋訓練點，並使用光譜角映射(Spectral angle mapping, SAM)關係當作過維度擴充(Band Expansion, BEP)條件之一，再應用限制能量最小化 (Constrained energy minimization, CEM) 凸顯出病灶分布位置，再使用歐蘇法(Otsu''s method)計算適合的閥值，最後以二值化結果呈現。
實驗結果發現，在合成影像上不同雜訊等級中的偵測情形中，執行SGA搭配SAM進行維度擴充後，將大幅提高偵測效率，與2013年張智凱所提出的ATGP搭配CEM組合方法以及與FCM分類方法多方比較之後，發現本實驗平均分類效果最佳。

In recent years, thanks to the advances of medical technology, brain disease is no longer a desperate end. Different diseases could affect gray-matter and white-matter change unpredictably, hence may cause brain lesion and brain degeneration differently. Therefore, how to segment the lesion efficiently and accurately is an important issue.
Fortunately, it was due to thousands of non-intrusive inspections that doctors don''t have to cut through the skin and muscle to confirm the lesion location. The most important and widely used non-intrusive inspections is using magnetic resonance imaging (MRI) technique. In this thesis we propose a SGA method for brain MRI tissue classification.
Classification algorithm generally requires a set of training samples to carry out in a supervised or an unsupervised manner depending upon how the training samples are produced a priori using prior knowledge or a posteriori obtained directly from the data. In this thesis, we present an unsupervised simplex growing algorithm (SGA) to find initial training sample. The initial training sample then combine with spectral angle mapping (SAM) technology to expand bands and regenerate a set of training samples. Then we apply constrained energy minimization (CEM) method to highlight the location of lesions. After all that work, we use Otsu''s method to calculate an appropriate threshold from the brain web image. The detection results that using SGA collocation with SAM method to expand bands, even if in the high noise surrounding, are better than the method using ATGP. Experiment results show that these approaches have greater promise in MR image classification than Chih-Kai Chang proposed thesis in 2013.

中文摘要 i
英文摘要 ii
目錄 iv
圖目錄 vi
表目錄 viii
第一章 緒論 1
第二章 背景 3
2.1 核磁共振造影 3
2.2 多頻譜影像處理技術 (Multi-spectral image processing technolgy) 4
2.3 機器學習 (Machine learning) 4
2.4 實驗條件 5
第三章 非監督式方法 9
3.1 自動目標物產生過程(Automatic Target Generation Process, ATGP) 9
3.2 單形體體積生長演算法(Simplex growing algorithm, SGA) 9
3.3 光譜角映射 (spectral angle mapper, SAM) 11
3.3.1 簡介 11
3.2.2 光譜角映射數學公式 11
3.4 維度擴充過程(Band expansion process, BEP) 12
3.5 限制能量最小化 (Constrained energy minimization approach, CEM) 12
3.6 Tanimoto Index 14
3.6.1 簡介 14
3.6.2 Tanimoto Index數學公式 14
3.7 歐蘇法(Otsu’s method) 14
3.7.1 簡介 14
3.7.2 演算法 15
3.8 正交子空間映射(Orthogonal subspace projection, OSP) 16
3.9模糊c-均數演算法(Fuzzy c-means clustering algorithm, FCM) 18
第四章 實驗 19
4.1 模擬大腦數據 19
4.2 SS+BEP+ATGP+SAM+CEM+ 2D Tanimoto Index+OSP 19
4.2.1 二維Tanimoto Index (2-Dimensional Tanimoto Index) [20]-[31] 20
4.2.2 實驗原理 20
4.3 SS+BEP+ATGP+ CEM+ FCM 20
4.4 SS+SGA+SAM+BEP+CEM+ Otsu’s method +OSP 21
4.5 實驗成效 24
4.5.1 第101組切片具0%雜訊與非均勻強度0% 24
4.5.2 第101組切片具1%雜訊與非均勻強度0% 25
4.5.3 第101組切片具3%雜訊與非均勻強度0% 26
4.5.4 第101組切片具5%雜訊與非均勻強度0% 27
4.5.5 第101組切片具1%雜訊與非均勻強度20% 28
4.5.6 第101組切片具3%雜訊與非均勻強度20% 29
4.5.7 第101組切片具5%雜訊與非均勻強度20% 30
4.5.8 量化分析 31
4.5.9 實際腦部分類 37
第五章 結論 38
參考文獻 39

[1] http://zh.wikipedia.org/wiki/
[2] http://www.wisegeek.org/what-are-white-matter-lesions.htm
[3] Y. J Chiou, H. M Chen, J. W Chai, C. C. C Chen, Y. C. Ouyang, W. C Su, C. W Yang, S. K Lee, C. I Chang, “Brian Tissue Classification Using Independent Vector Analysis(IVA) For Magnetic Resonance Image” IEEE International Conference on Bioinformatics and Bioengineering, 52: 324-329, 2009
[4] http://www.bic.mni.mcgill.ca/brainweb/
[5] C. I. Chang, “Automatic Spectral Target Recognition in Hyperspectral Imagery, ” Manuscript received August 10, 2001; revised December 10, 2002; released for publication October 4, 2003.
[6] C. I Chang, S.Y Chen, H. C Li, Chinsu Lin, “A comparative study among ATGP,VCA and SGA for finding endmembers in hyperspectral imahery” J-STARS, 2014
[7] C. I Chang, C.C Wu, W.M Liu, Y.C Ouyang “A New Growing Method for Simplex-Based Endmember Extraction Algorithm” IEEE Transactions on Geoscience And Remote Sensing, vol. 44, No.10,October 2006
[8] C.I.Chang, Hyperspectral data processing：Algorithm Design and Analysis, John Wiley & Sons, Inc., Publication.
[9] Boardman, 1992, SIPS User’s Guide Spectral Image Processing System, Version 1.2, Center for the Study of Earth from Space, Boulder.
[10] F. A. Kruse, A.B Lefkoff, J.W. Boardman, Heiedbrecht, A.T., P.J. Barloon, Goetz, A.F.H.,1992. The Spectral Image Processing System (SIPS) - Software for Integrated Analysis of AVIRIS DataSummaries of the 4th Annual JPL Airborne Geoscience Workshop, JPL Pub-92-14, AVIRIS Workshop. Jet Propulsion Laboratory, Pasadena, CA, pp. 23-25.
[11] J. C. Harsanyi, “Detection and classification of subpixel spectral signatures in hyperspectral image sequences,” Ph.D. dissertation, Dept, Elect. Eng., Univ. Maryland Baltimore County, Baltimore, MD, 1993.
[12] J. C. Harsanyi,W. Farrand, and C.I. Chang, “Detection of subpixel spectral signatures in hyperspectral image sequences,” in Proc. American Society of Photogrammetry & Remote Sensing Annu. Meeting, Reno, 1994, pp. 236–247.
[13] R. S. Resmini, M.E. Kappus, W. S. Aldrich, J.C. Harsanyi, and M. Anderson, “Mineral mapping with Hyperspectral Digital Imagery Collection Experiment (HYDICE) sensor data at Cuprite, Nevada, U.S.A.,” Int. J. Remote Sensing, vol. 18, no. 17, pp. 1553–1570, 1997.
[14] W. Farrand and J.C. Harsanyi, “Mapping the distribution of mine tailing in the Coeur d’Alene river valley, Idaho, through the use of constrained energy minimization technique,” Remote Sensing of Environment, vol. 59, pp. 64–76, 1997.
[15] S. Theodoridis and K. Koutroumbas, Pattern Recognition,Academic Press, 1999, p. 366
[16] Nobuyuki Otsu. “A Threshold Selection Method from Gray-Level Histograms,” IEEE Transactions on System, Mam, and Cybernetics, Vol. SMC-9, No. 1, 1979, pp. 62-66.
[17] http://www.itrc.narl.org.tw/Publication/Newsletter/no84/p14.php
[18] J. C. Harsanyi, and C. I. Chang (1994) Hyperspectral image classification and dimensionality reduction: An orthogonal subspace projection. IEEE Transactions on Geoscience and Remote Sensing, 32 (1994), 779–785.
[19] Y.C Ouyang, C.K Chang, 2013 “Unsupervised MS lesion detection and classification method on brain MRI images”, master’s thesis. N.C.H.U., Taichung
[20] C. I.Chang(張建禕), E. Wong ”TWO-DIMENSIONAL TANIMOTO INDEX FOR CONTINUOUS DECISION MADE IMAGE CLASSFICATION,” conference on Computer Vision Graphics, and Image Processing, 2012”
[21] S. Theodoridis and K. Koutroumbas, Pattern Recognition,Academic Press, 1999, p. 366
[22] Duda and Hart, Pattern Classification and Scene Analysis, 1973.
[23] C. I. Chang, “Multiple-parameter receiver operating characteristic analysis for signal detection and classification,” IEEE Sensors Journal, vol. 10, no. 3, pp.423-442, March 2010.
[24] www.bic.mni.mcgill.ca/ brainweb/
[25] http://www.fil.ion.ucl.ac.uk/spm/software/SPM8/
[26] S. M. Smith, “Fast Robust Automated Brain Extraction,” Human Brain Mapping, vol. 17, pp. 143–155, 2002.
[27] Y. Zhang, M. Brady, and S. Smith. “Segmentation of Brain MR Images Through a Hidden Markov Random Field Model and the Expectation-Maximization Algorithm,” IEEE Trans. Medical Imaging, vol. 20, pp. 45-57, 2001.
[28] S. M. Smith, M. Jenkinson, M. W. Woolrich, C.F. Beckmann, T. E.J. Behrens, H. Johansen-Berg, P. R. Bannister, M. D Luca, I. Drobnjak, D. E. Flitney, R. K. Niazy, J. Saunders, J. Vickers, Y. Zhang, N. D. Stefano, J. M. Brady, and P. M. Matthews. “Advances in functional and structural MR image analysis and implementation as FSL,” NeuroImage, vol. 23: S208–S219, 2004.
[29] http://www.fmrib.ox.ac.uk/fsl/
[30] E. Wong, Partial Volume Estimation of Magnetic Resonance Image Using Linear Spectral Mixing Analysis, Ph.D. dissertation, Department of Computer Science and Electrical Engineering, University of Maryland, Baltimore County, May 2010.
[31] Boardman, 1992, SIPS User’s Guide Spectral Image Processing System, Version 1.2, Center for the Study of Earth from Space, Boulder.
[32] http://mirlab.org/jang/books/dcpr/index.asp