臺灣博碩士論文加值系統

由於功能性磁振影像具有空間及時間的高解析度，以及它不需使用具有放射活性之顯影劑等特性，使得它在未來數十年成為科學研究與臨床應用最具發展性之利器。功能性磁振影像提供了在大腦活化區對於帶氧血紅素與不帶氧血紅素比值之直接觀察。經由氧氣的供應，消耗以及新陳代謝率等指標，我們得到對於大腦功能性反應之間接性推測。
由於功能性磁振影像的低訊雜比，影像後處理在凸顯功能性反應的程序中，仍為不可或缺的步驟。傳統的影像後處理，大多依賴具閥值決定能力之統計方法。例如：統計參數圖譜法及相關係數分析法等。然而，此種以統計為基礎的分析法，需要具備有關實驗模式的資訊或者有關大腦反應類型之假設。可是，現實上這些並非每次皆可得。是故，此類分析偏差，可能造成磁振影像資料解讀的困難及誤導。因此，資料導向的分析法，對功能性磁振影像而言，特別是我們若要研究較複雜之大腦功能就顯得相當重要了。
本論文之主要貢獻在於我們提出並發展了結合空間及時域的功能性影像之分析法。而這兩者皆根植於非監督式的柯氏特徵圖譜神經網路與C平均模糊邏輯演算法的接續與合作。柯氏神經網路是一種自我組織之特徵圖譜，它能將磁振影像中的特徵值依其彼此之相似程度而分類成群集。C平均模糊邏輯也從事件特徵值分類，它將一個輸入向量，歸至每一個輸出向量的機率的歸屬函數計算出來。柯氏特徵圖譜的計算力很強，且進行資料減量及濃縮的效率很高，因此我們用它做第一次通過的處理演算法，主要是著眼於其高效能的計算與使用大量群集時，所可達成之高敏感度。C平均模糊邏輯則用在第二階段的處理中，它所擅長的是分出相似、曖昧、混沌不清的特徵值，它被用來合併及重組這些特徵值，使它們分配到小數目的群集去，接著它把每個畫素依其所屬之分類進行標示。本論文所提出的空間分析策略，即是自動化的影像與組織之分割，這有助於區分在腦實質內或腦血管中的反應。時間分析策略則在辨認反應模式。本方法成功地找到用任意運動模式實驗下所產生的反應模式。不論是神經網路或模糊邏輯，它們都是經過迭代反覆運算過程而達到群集與分類的目的。它們一樣會面臨初始值設定，最佳化群集數…等問題。這些在論文之內都有討論。
功能性磁振影像技術在基礎神經科學的研究與臨床神經學的應用的遠景可謂相當豐富。我們進行了動物電刺激反應的實驗並探討正常鼠之體位感覺的生理功能以及中風鼠其缺血性中風對體位感覺功能的影響與恢復狀態。人體方面的實驗則涵括了運動誘發反應、視覺刺激以及電刺激之體位感覺反應。將來更進一步的整合多重模式的腦部功能性影像及不同模式之間的對位。例如功能性磁振影像，正子斷層掃瞄，腦電波及腦磁波的合併應用，期能發展出探討人類大腦奧秘的有力詳實的鑰匙，在這同時也開發出臨床神經學與神經外科之重要輔助工具。

Functional magnetic resonance imaging (fMRI) is bound to be a promising tool for scientific research and clinical practice in the coming decades for its relative high spatial and temporal resolution and its lacking potential hazard from radioactive contrast medium. It provides direct observation of the changes in oxyhemoglobin / deoxyhemoglobin ratios, i.e., the blood oxygen level dependent (BOLD) effect, in the region of brain activation. Thus, it makes available an indirect reference of the functional response of the brain as indexed from oxygen supply, consumption and metabolic rate.
Since the signal-to-noise ratio is low in fMRI, image post-processing is always necessary to identify the functional response. The conventional image post-processing methods often rely on statistics capable of threshold determining, such as statistic parametric mapping or correlation coefficient analysis. However, the statistics-based method requires knowledge of the imposed experiment paradigm or an assumption on pattern of brain response. Nevertheless, this may not always be available. Hence, biased results from such analyses could be produced, which make the interpretation of fMRI data difficult or misleading.
Therefore, methods for data-driven analysis in fMRI become crucial especially when we intend to investigate some sophisticate brain function. The major contribution of this thesis is that we proposed and developed strategies combining spatial and temporal analyses both based on unsupervised algorithms cascading Kohonen-feature-map (KFM) neural network and fuzzy C-means. Kohonen network is a self-organizing feature map, which is capable of identifying characteristic features from the MR images and classifying those pixels with strong resemblance into clusters. Fuzzy C-means does feature classification by assigning a membership function describing the probability of one input vector to go to each output cluster. KFM has high computation power and good efficacy for data reduction. Therefore, we used it as a first-pass-processing algorithm aiming at computation efficiency and high sensitivity by using many clusters. Fuzzy C-means, used in the second stage of processing, being good at separating similar and ambiguous features, is used to merge and to reorganize those features into clusters of small number. It then labels each pixel to its belonging classification. The spatial strategy, proposed in this thesis, performed automated tissue segmentation, which helped to separate response in the parenchyma form the vessel space. The temporal strategy identified pattern of response, which were carried out by an arbitrary paradigm, with satisfactory results. Either neural network or fuzzy algorithm capable of clustering and classifying through iterative processing suffers problems of initial value condition, optimal number of the clusters …etc. These are all examined and discussed in this thesis.
Prospects of fMRI application in basic neuroscience research and clinical neurology are rather futile. We conducted animal studies on the electric stimulation in both healthy and stroke rats to explore the physiology of the somatosensory function and the ischemic effect and recovery on such function. Human studies of motor evoked response, visual stimulation, and electric shock induced somatosensory responses were all performed. In the future, further integration and registration of multi-modal functional images such as fMRI, PET, EEG and MEG will formulate a comprehensive and powerful key in exploring secretes of the human brain. At the same time, it constructs a great assistance for clinical neurology and neurosurgery.

COGVER
Contents
CHINESE ABSTRACT
ENGLISH ABSTRACT
1.INTRODUCTION
1.1 THE IMPORTANCE OF FMRI AS A TOOL FOR CLINICAL AND BASIC NEUROSCIENCE RESEARCH
1.2 PROBLEMS TO BE SOLVED IN FMRI
1.3 MOTIVATION
1.4 BACKGROUND
1.4.1 The Development of MRI and fMRI
1.4.2 Principle of fMRI
1.5 PHYSIOLOGY AND NEURAL MECHANISM OF HUMAN BRAIN ACTIVATION
1.5.1 Functional Anatomy and Nwurophysiology of Visual Activation
1.5.2 Functional Anatomy and Nwurophysiology of Motor Activation
1.5.3 Functional Anatomy, Neurophysiology of Somatosensory Activation
1.6 THE ANALYSIS OF FMRI SIGNAL
1.6.1 Methods of Post Processing
1.6.2 Contemporary Method of Analysis
1.6.3 Unknown Patterms of Response to be Identified in terms of Spatia Distribution and Temporal Profiles
1.7 PULSE SEQUENCE ISSUES
1.8 THE OPERATIONAL COMPONENTS OF FUNCTIONAL
1.8.1 Image Data Analysis
1.8.2 Realignment
1.8.3 Spatial Normalization
1.8.4 spatial Smoothing
1.8.5 Statistical Analysis
1.8.6 Statistical Inference
1.9 SUMMARY
2. METHODS
2.1 IMAGE ACQUISITION
2.1.1 GRE0Flash
2.1.2 EPI
2.2 IMAGE PREPROCESSING
2.2.1 Resistration
2.2.2 Image Enhancement and Background Canceling
2.3 STATISTIC ANALYSIS
2.3.1 correlation Coefficient Analysis
2.3.2 Statistic Parametric Mapping
2.4 MATCHED FILTERS
2.5 ROC ANALYSIS ON STATISTIC ANALYSIS
2.6 AUTOMATED ANALYSIS
2.6.1 Kohonen Feature Map
2.6.2 Fuazy C0Means
2.6.3 The Combination of Kohonen Feature Map and Fuzzy C0Means
2.6.4 Spatial Strategy0Automated Segmentation as an Aid for Tissue Specific Analysis of the Functional Response
2.6.5 Temporal Strategy0Identification of Unknown Pattern of Response in the Brain Activation
2.7 APPLICATION
2.7.1 Animal Studies
2.7.2 Physiological Study
2.7.3 Animal Focal Ischemic Model Study MCA Ligation
2.7.4 Histomorphological/Immunochemical Study
2.7.5 Human Studies
2.7.6 Physiology Study
2.7.7 Clinical Study
3. RESULTS
3.1 RESULTS OF THE STATISTICAL ANALYSIS ON FMRI DATA
3.1.1 Results of ROC Analysis
3.2 AUTOMATED TISSUE SEGMENTATION OF MR IMAGES
3.3 IDENTIFICATION OF UNKONWN RESPONSE CURVES
3.3.1 Regular Paradigm Experiment
3.3.2 Arbitrary Paradigm Experiment
3.4 RESEARCH OF CLINICAL AND BASIC NEUROSCIENCE
3.4.1 Sensory Evoked Respinse in Rat with Electric Stimulation
3.4.2 Effect of Ischemia on Sensory Evoked Response in Brain Tissue A Focal Cerebral Ischemic Model of Animal and fMRI
3.4.3 Histomorphological Correlation of the Cerebral Infarcts with MRI and fMRI
3.4.4 Motor Response of Human Subjects
3.4.5 Visual Response of Human Subjects
3.4.6 Cross Modality Study of the fMRI and EEG (Evoked Potentials)
3.4.7 Application of Patients with Cerebral Ishemic Infarcts
4. DISCUSSION
4.1 THE SPATIAL AND TEMPORAL STRATEGIES ON ANALYSES OF FMRI SIGNALS WITH KOHONEN NEURAL NET AND FUZZY C0MEANS
4.1.1 Automated Tissue Segmentation Provides Further Spatial Information for Neuroscientists to do analyses on fMRI Data for the Interpretation of Response and Mechanism with Brain Activation
4.1.2 The Identification of Unknown Patterns of Response in fMRI Offer Further Insights for the Temporal Profiles and Events of the Brain Mechanism
4.1.3 Inter0Modality Study of Functional Brain Mapping Is Necessary for Comprehension of Brain Function
4.2 INTERPRETATION AND BIOLOGICAL SIGNIFICANCE OF FMRI SIGNAL
4.2.1 Inflow Versus BOLD Effect
4.2.2 Intrinsic Paramagnetic Substance Indicating Cerebral Oxygenation and Implying Metabolism?
4.2.3 Effect of the Task Complexity
4.3 PROBLEMS WITH BASIC NEUROSCIENCE RESEARCH
5. CONCLUSIONS
5.1 SPATIAL AND TEMPORAL STRATEGIES FOR FMRI ANALYSES0UNSUPERVISED METHODS COMBINING ARTIFICIAL NEURAL NETWORK AND FUZZY LOGIC''S AS AN IMPORTANT TOOL FOR SCIENTIFIC RESEARCH AND CLINICAL PRACTICE
5.2 FUTURE WORK IN TECHNOLOGICAL ASPECTS OF FMRI
5.3 FUTURE WORK IN BASIC NWUROSCIENCE RESEARCH
5.3.1 Identification and Interpretation of Physiological Response of Different Domain in terms of Tissue Specificity and Individual Temporal Profile
5.4 FUTURE WORK IN CLINICAL NEUROSCIENCE RESEARCH
5.4.1 Standardized Paradigm for Clinical Application
5.4.2 Normal Data Bank for Different Modalities of Brain Activation
5.4.3 Issus of Quantitative Analysis of fMRI
5.4.4 Inter0Modal Correlation Study and Multi0Modal Registration
OTHERS

1. Ogawa S, Lee TM, Kay AR, Tank DW. Brain magnetic resonance imaging with contrast dependent on blood oxygenation. Proc Natl Acad Sci USA 1992;87:9868—72.
2. Ogawa S, Lee TM, Nayak AS, Glynn P. Oxygen-sensitive contrast in magnetic resonance image of rodent brain at high magnetic fields. Magn Reson Med 1990;14:68—78.
3. Kwong KK, Belliveau JW, Chesler DA, et al. Dynamic magnetic imaging of human brain activity during primary sensory stimulation. Proc Natl Acad Sci USA 1992;89:5675—9.
4. Duyn JH, Moonen CTW, van Yperen GH, de Boer RW, Luyten PR. Inflow versus deoxyhemoglobin effects in BOLD functional MRI using gradient echoes at 1.5 T. NMR Biomed 1994;7:83—8.
5. Segbearth C, Belle V, Delon C, et al. Functional MRI of the human brain: predominance of signals from extracerebral veins. Neuroreport 1994;5:831—6.
6. Kim SG, Ugurbil K. Comparison of blood oxygenation and cerebral blood flow effects in fMRI: Estimation of relative oxygen consumption change. Magn Reson Med 1997;38:59—65.
7. Gao JH, Miller I, Lai S, Xiong J, Fox PT. Quantitative assessment of blood inflow effects in functional MRI signals. Magn Reson Med 1996;36:314—9.
8. Gati JS. Menon RS, Ugurbil K, Rutt BK. Experimental determination of the BOLD field strength dependence in vessels and tissue. Magn Reson Med 1996;38:296—302.
9. Lauterbur PC. Image formation by induced local interactions: examples employing nuclear magnetic resonance. Nature 1973;242:190.
10. Sobel DF, Gallen CC, Swartz BJ, et al. Locating the central sulcus: a comparison of MR anatomic and magnetoencephalographic functional methods. Am J Neuroradiol 1993;14:915.
11. Belliveau JW, Kwong KK, Kennedy DN, et al. Magnetic resonance imaging mapping of brain function: human visual cortex. Invest Radiol 1992;27:559.
12. Belliveau JW, Rosen BR, Kantor HL, et al. Functional cerebral imaging by susceptibility contrast NMR. Magn Reson Med 1990;14:538.
13. Bryan JM, Lowe MJ, Turski PA: Functional Magnetic Resonance Imaging. In: Stark DD, Bradley WG, eds. Magnetic Resonance Imaging. 3rd ed. St. Louis, Mo. Mosby, 1999:1555-74.
14. Belliveau JW, Kennedy DN, McKinstry RC et al: Funcitonal mapping of the human visual cortex by magnetic resonance imaging, Science 254:716-9, 1991.
15. Detre JA, Leigh JS, Williams DS, et al. Perfusion imaging. Magn Reson Med 1992;23:37.
16. Williams DS, Detre JA, Leigh JS, et al. Magnetic resonance of perfusion using spin inversion of arterial water. Proc Natl Acad Sci USA 1992;89:212.
17. Thulborn KWJ, Matthews P et al. Oxygenation dependence of the transverse relaxation time of water protons in whole blood at high field. Biochem Biophys Acta 1982;714:265.
18. Ogawa S, Lee T. Magnetic resonance imaging of blood vessels at high fields: in vivo and in vitro measurements and image simulation. Magn Reson Med 1990;16:9.
19. Fox PT, Raichle ME. Focal physiological uncoupling of cerebral blood flow and oxidative mechanism during somatosensory stimulation in human subjects. Proc Natl Acad Sci USA 1986;83:1140—4.
20. Deyoe E, Bandettini P, Neitz J et al. Functional magnetic resonance imaging (fMRI) of the human brain. J Neurosci Methods 1994;54:171.
21. Zeki S. The functional organization of projection from striate to prestriate visual cortex in the rhesus monkey. Cold Spring Harbour Symp. Quant. Biol. 1975;40:591-600.
22. Livingstone MS, Hubel DH. Specificity of intrinsic connections in primate primary visual cortex. J. Neurosci 1984;4:2830-2835.
23. Zeki S, Watson JDG., Lueck CJ, Friston KJ, Kennard C, Frackowiak RSJ. A direct demonstration of functional specialization in human visual cortex. J. Neurosci 1991;11:641-649.
24. Leyton ASF, Sherrington CS. Observation on the excitable cortex of the chimpanzee, orang-utang, and gorilla. QJ Exp Physiol 11:135-222, 1971.
25. Penfield W, Rasmussen T. the cerebral cortex of man: A clinical study of localization of function. New York: Macmillan 1950.
26. Bandettini PA, Jesmanowicz A, Wong EC, Hyde JS. Processing strategies for time-course data sets in functional MRI of the human brain. Magn Reson Med 1993;30:161-173.
27. Friston KJ, Jezzard P, Turner R. Analysis of functional MRI time-series. Human brain mapping 1994;1:153-171.
28. Khosla D, Singh M, Patel P. Principle component analysis to detect fMRI activity. Proc ISMRM 4th Ann meeting, New York, 1996, p.1797.
29. Talairach P, Tournoux, J. A stereotactic coplanar atlas of the human brain. Thieme. Stuttgart, 1998.
30. Fox PT, Perlmutter JS, Raichle ME. A stereotactic method of anatomical localization for position emission tomography. J. Comput Assist. Tomogr 1985;9:141-153.
31. Frackowiak RSJ, Friston KJ, Frith CD, Dolan RJ, Mazziotta JC. Human Brain Function. Chapter 2. Academic Press. 1997.
32. Woods RP, Cherry SR, Mazziotta JC. Rapid automated algorithm for aligning and reslicing PET images. J Comput Assist Tomogr 1992;16:620—33.
33. Jackson EF, Narayana PA, Falconer JC. Reproducibility of Nonparametric Feature Map Segmentation for Determination of Normal Human Intracranial Volumes with MR Imaging Data. J Magn Reson Imaging 1994;4(5): 692-700.
34. Zurada JM. Introduction to artificial neural systems. Chap. 7. West Publishing Co., 1992.
35. Kohonen T. Self-Organization and Associative Memory. Springer-Verlag, New York, 1984.
36. Yeh IC. Artificial Neural Network Model Application and Implementation. Scholars Book Co., Taipei, 1993.
37. Wasserman PD. Neural Computing Theory and Practice, ANZA Research Inc., New York, 1989.
38. Hall LO, Bensaid AM, Clarke LP, Velthuizen RP, Silbiger MS, Bezdek JC. A comparison of neural network and fuzzy clustering techniques in segmenting magnetic resonance images of the brain. IEEE Trans Neural Netw 1992;3(5):672-682.
39. Clark MC, Hall LO, Goldgof DB, Clarke LP, Velthuizen RP, Silbiger MS. MRI Segmentation Using Fuzzy Clustering Techniques. IEEE Eng Med Biol Mag 1994;13(5):730-742.
40. Phillips WE, Velthuizen RP, Phuphanich S, Hall LO, Clarke LP, Silbiger ML. Application of Fuzzy c-Means Segmentation Technique For Tissue Differentiation in MR Images of a Hemorrhagic Glioblastoma Multiforme. Magn Reson Imag 1995;13(2):277-290.
41. Cannon RL, Dave JV, Bezdek JC: Efficient Implementation of the Fuzzy c-Means Clustering Algorithms. IEEE Trans on Pattern Anal Mach Intell 8: 248-55, 1986
42. Bezdek JC, Tsao EC, Pal NR: Fuzzy Kohonen Clustering Networks. Proc. IEEE Intl. Conf. Fuzzy Systems, San Diego, USA, 1992;1035-1043.
43. Chiu MJ, Lin CC, Chuang KH, et al. Visualization and feature extraction of brain information: Automated segmentation on magnetic resonance images and functional magnetic resonance images. Submitted to IEEE trans Inform Technol in Biomed 2000.
44. Pal NR, Bezdek JC, Tsao EC-K. Generalized clustering networks and Kohenen''s self-organizing scheme. IEEE Trans. Neural Network 1993;4:549-556.
45. Overgaard K, Sereghy T, Boysen G, Pedersen H, Diemer NH. Reduction of infarct volume and mortality by thrombolysis in a rat embolic stroke model. Stroke 1992;23:1167-1174.
46. Baumgertner R, Windischberger C, Moser E. Quantification in functional magnetic resonance imaging: Fuzzy clustering vs. correlation analysis. Magn Reson Imag 1998;16(2):115-125.
47. Golay X, Kollias S, Stoll G, Meier D, Valavanis A, Boesiger P. A new correlation-based fuzzy logic clustering algorithm for fMRI. Magn Reson Med 1998;40:249-260.
48. Chiu MJ, Chuang KH, Chen JH, Huang KM. Response of the human motor cortex-an application of the functional MRI. Biomed Eng Appl Basis Comm 1998;10:326-331.
49. Clarke LP, Velthuizen RP, Camacho MA, et al. MRI segmentation: methods and applications. Magn Reson Imag 1995;13: 343-68.
50. Bomans M, Hohne KH, Tiede U, et al. 3D segmentation of MR images of the head for 3-D display. IEEE Trans Med Img 1990;9: 177-83.
51. Bezdek JC, Hall LO, Clarke LP. Review of MR image segmentation techniques using pattern recognition. Med Phys 1993;20: 1033-48.
52. Chuang KH. Identification of fMRI signal using Fuzzy Neural Network. MA Thesis. Taipei: National Taiwan University. 1998.
53. Chuang KH, Chiu MJ, Lin CC, Chen JH. Finding different physiological signal sources of fMRI data using paradigm-free Kohenen clustering network and fuzzy c-means. In Proc. ESMRMB 15th Annu. Meeting. Geneva, Switzerland. 1998;143-144.
54. Kohonen T. Self-organizing maps. Springer-Verlag, New York, 1995.
55. Kim SG, Tsekos NV. Perfusion imaging by a flow-sensitive alternating inversion recovery (Fair) technique: Application to functional brain imaging. Magn Reson Med 1997;37:425-435.
56. Constable RT, McCarthy G, Allison T, Anderson AW, Gore JC. Functional brain imaging at 1.5 T using conventional gradient echo MR iamging technique. Magn Reson Imag. 1993;11:451—9.
57. Price RR, Lee H, Welch L, et al. Fucntional MR iamging assessment of fine motor function. Proc 4th Ann Meeting of ISMRM 1996, April, New York, USA.
58. Saunders JS, Wowk B, Scarth G, et al. Thinking, thinking harder — the fMRI of graded cognitive/motor activation. Proc 3th Ann Meeting of ISMRM 1995, April. Nice, France.
59. Rao SM, Biner JR, Bandettini PA, Hammeke TA, et al. Functional magnetic resonsnace imaging of complex human movements. Neurology 1993;43:2311—8.
60. Hatazawa J, Ito M, Matsuzawa T, Ido T, Watanuki SI. Measurement of the ratio of cerebral oxygen consumption to glucose utilization by positron emission tomography: Its consistency with the values determined by the Kety-Schmidt method in normal volunteers. J Cereb Blood Flow Metab 1998;8:426—32.
61. Pantano P, Baron J-C, Lebrun-Grandie P, Duquesnoy N, Bousser M-G, Comar D. Regional cerebral blood flow and oxygen consumption in human aging. Stroke 1984;15:635—42.
62. Colebatch JG, Deiber M-P, Passingham RE, Friston KJ, Frackowiak RSJ. Regional cerebral blood flow during voluntary arm and hand movements in human subjects. J Neurophysiol 1991;65:1392—1401.
63. Roland PE, Meyer E, Shibasaki T, Yamamoto YL, Thompson CJ. Regional cerebral blood flow changes in cortex and basal ganglia during voluntary movements in normal human volunteers. J Neurophysiol 1982;40:467—80.
64. Grafton ST, Woods RP, Mazziotta JC, Phelps ME. Somatotopic mapping of the primary motor cortex in humans: activation studies with cerebral blood flow and positron emission tomography. J Neurophysiol 1991;66:735—43.
65. Lahti KM, Ferris CF, Li F, Sotak CH, King JA. Comparison of evoked cortical activity in conscious and propofol-anesthetized rats using functional MRI. Magn Reson Med. 1999;41(2):412-416.
66. Antognini JF, Buonocore MH, Disbrow EA, Carstens E. Isoflurane anesthesia blunts cerebral responses to noxious and innocuous stimuli: a fMRI study. Life Sciences. 1997:61(24):PL 349-354.
67. Chen CK, Chiueh TD, Chen JH. Active cancellation system of acoustic noise in MR imaging. IEEE Trans Biomed Eng. 1999:46;186-91.