以貝氏定理為基礎之神經疾病磁振造影影像電腦輔助評估系統
18
0
0
全文
(2)
(3) 以貝氏定理為基礎之神經疾病磁振造影影像電腦輔助評估系統 Computer-aided MRI Evaluation of Neurological Diseases based on Bayes' Theorem. 研 究 生：張雅婷. Student：Ya-Ting Chang. 指導教授：陳永昇. Advisor：Yong-Sheng Chen. 國 立 交 通 大 學 資 訊 科 學 與 工 程 研 究 所 碩 士 論 文. A Thesis Submitted to Institute of Computer Science and Engineering College of Computer Science National Chiao Tung University in partial Fulfillment of the Requirements for the Degree of Master in. Computer Science August 2007 Hsinchu, Taiwan, Republic of China. 中華民國九十六年八月.
(4)
(5) 摘. 要. 近年來，由於醫學影像分析技術的蓬勃發展，電腦輔助診斷系統也隨之成為 研究的潮流。以往的診斷是依賴醫師們的專業判斷，但這樣的判斷會受限於醫師 主觀的判斷，且較小不易被肉眼察覺的部分容易被忽略。此外，受測者需花費許 多時間等待結果。因此，建造一套簡單且具高正確率的電腦輔助診斷系統可以客 觀且省時的方式，提供醫師及受測者們參考的指標。目前已存在許多輔助診斷系 統，但多數的系統僅提供絕對性的參考指標。而在我們提出的方法中，我們以機 率值的方式呈現估測的結果，提供醫師及受測者一個具有程度差異的相對性指 標。 在本研究中，我們將圖形識別(pattern recognition)的技術應用在建立輔助系 統上。整個系統是由數個平行的分類模組所組成，且每一個分類模組單就針對一 種特定疾病作分析。每一個分類模組的建立都需經由兩個步驟：特徵擷取(feature selection and extraction)與分類(classification)。首先，透過以體素為基礎的型態計 量學(voxel-based morphometry, VBM)，找出某一特定疾病病患與正常人的腦部結 構差異所在，並將這些具有鑑別力的特徵選取出來。再者，應用主要成分分析 (principal component analysis)技術找到最合適的資料表示方式，並採用兩種方式 篩選合適的主軸建構分類空間，分別稱為以變異量為基礎的主軸挑選方法 (variance-based PC selection)及以鑑別力為基礎的主軸挑選方法(significant-based PC selection)。最後應用貝氏定理(Bayes’ Theorem)配合非參數密度估測方法 (nonparametric density estimation – Parzen Windows)，估測受測者罹患某種特定疾 病的機率。 我們將此分類架構應用在脊髓小腦運動失調症(spinocerebellar ataxia type III, SCA3)及躁鬱症(bipolar disorder, BD)的研究中，且各用兩種主軸挑選方式皆各自 建構對應的分類器。我們發現採用以鑑別力為基礎的主軸挑選方法所建構的分類 器，能達到較好的系統效能，且其呈現的結果也較為合理一致。. 壹.
(6) 貳.
(7) 誌. 謝. 在碩士生涯中要感謝很多人給予的支持與鼓勵。首先是兩位指導老師，陳永 昇老師與陳麗芬老師，在研究的路途上，感謝你們對我的指導，才使我能有現在 的成果；也感謝實驗室裡所有的成員，不僅是研究上相互交流，在娛樂生活上更 是玩樂的好伙伴，很喜歡跟大家一起打球、出遊的日子。此外也要相當感謝一起 對畢業奮鬥的戰友，讓我在沮喪的時候有人可以聽我吐苦水，而發洩完後有動力 重新出發。 最後要感謝十分重要的家人，因為一路上有你們的支持，我才可以無憂無慮 的完成碩士學位，媽媽總是叮嚀我不要太累，姊姊會在寒冷的冬天裡給我一件溫 暖的外套，而爸爸雖然老是搞不懂我在做什麼，但會陪我一起熬夜，尤其當我在 研究不順遂時，你們總能給予我支持與鼓勵，真的很溫暖，謝謝。. 參.
(8) 肆.
(9) Computer-aided MRI Evaluation of Neurological Diseases based on Bayes’ Theorem A thesis presented by. Ya-Ting Chang to. Institute of Computer Science and Engineering College of Computer Science in partial fulfillment of the requirements for the degree of Master in the subject of. Computer Science National Chiao Tung University Hsinchu, Taiwan 2007.
(10) Computer-aided MRI Evaluation of Neurological Diseases based on Bayes’ Theorem. Copyright © 2007 by Ya-Ting Chang.
(11) Abstract. Recently, the study of computer-aided diagnosis (CAD) becomes a trend of biomedical signal processing due to developments from medical image analysis technology. In the past, a diagnosis depends on doctors’ judgments, is subjective to physicians and costs much time for subjects to get results. Moreover, subtle differences which reveal potential danger may be invisible to human eyes. Thus, a simple CAD system with high correct accuracy can supply an index sign for physicians and subjects in an objective and convenient way. Most of existent systems, however, provide an absolute prediction on a test subject. It means that the answer would be either yes or no. Therefore, we propose a probabilistic approach to tell doctors and test subjects probabilistic predictions which show the difference of degree. In this thesis, we construct a computer-aided MRI evaluation system with statistical pattern recognition technology. The entire system is parallelly composed of several disease classification models and each classification model is aimed at classifying a particular disease. For each model, there are two processes: feature selection and extraction, and classification. Initially, locations where reveal significant anatomical discrepancy discovered by a voxel-based morphometric analysis (VBM) are picked out as distinguishable features for classification. Moreover, principal component analysis (PCA) is applied to find proper representations for those found features and some applicable PCs are chosen to establish a good classification space by two principal component (PC) selection methods. One is named as variance-based PC selection method and the other is significant-based PC selection method. Finally, the classification model predicts the possibility of a test subject to sicken with a particular disease by using Bayes’ Theorem and a nonparametric density estimation, Parzen windows. Our proposed classification framework was applied on spinocerebellar ataxia type III (SCA3) and bipolar disorder (BD) and two corresponding classification models were established separately. Both of two PC selection methods were used in each model. Thus, i.
(12) there were two distinct classifiers in a model. In our experiments, we found that a classifier with significant-based PC selection method not only achieves a better performance but also has a more consistent result.. ii.
(13) Contents List of Figures. v. List of Tables. vii. 1. 2. Introduction. 1. 1.1. Brain Structures and Magnetic Resonance Imaging . . . . . . . . . . . . .. 2. 1.2. Statistical Pattern Recognition . . . . . . . . . . . . . . . . . . . . . . . .. 6. 1.3. Computer-Aided Diagnosis . . . . . . . . . . . . . . . . . . . . . . . . . .. 8. 1.4. Thesis Scope and Organization . . . . . . . . . . . . . . . . . . . . . . . . 10. Feature Selection and Extraction 2.1. 2.2. 3. 15. Voxel-Based Morphometry . . . . . . . . . . . . . . . . . . . . . . . . . . 16 2.1.1. Introduction to VBM . . . . . . . . . . . . . . . . . . . . . . . . . 16. 2.1.2. Optimized VBM protocol . . . . . . . . . . . . . . . . . . . . . . 20. 2.1.3. Implementation of VBM . . . . . . . . . . . . . . . . . . . . . . . 24. Principal Component Analysis . . . . . . . . . . . . . . . . . . . . . . . . 28 2.2.1. Introduction to PCA . . . . . . . . . . . . . . . . . . . . . . . . . 28. 2.2.2. Feature Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . 32. Classification. 35. 3.1. Framework of Computer-Aided Diagnosis System . . . . . . . . . . . . . . 36. 3.2. Bayes’ Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41. 3.3. Parzen Windows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43. 3.4. Accuracy Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 iii.
(14) 4. 5. 6. Experiment Results. 51. 4.1. Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52. 4.2. Structural Analysis for Patients Suffering Spinocerebellar Ataxia Type 3 . . 53. 4.3. Structural Analysis for Patients Suffering Bipolar Disorder . . . . . . . . . 60. 4.4. Experiments on Diagnosis System . . . . . . . . . . . . . . . . . . . . . . 67 4.4.1. Results of SCA3 Diagnosis System . . . . . . . . . . . . . . . . . 68. 4.4.2. Results of BD Diagnosis System . . . . . . . . . . . . . . . . . . . 74. Discussion. 83. 5.1. Why not using whole voxels as features in classification . . . . . . . . . . . 84. 5.2. Influences of window sizes in Parzen-window approach . . . . . . . . . . . 86. 5.3. Comparisons between variance-based PC selection and significant-based PC selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91. 5.4. Cross-group testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93. Conclusions. 99. Bibliography. 105. iv.
(15) List of Figures 1.1. Main human brain structures . . . . . . . . . . . . . . . . . . . . . . . . .. 3. 1.2. Distributions of GM, WM and CSF in the brain . . . . . . . . . . . . . . .. 4. 1.3. Simple flowchart of magnetic resonance imaging . . . . . . . . . . . . . .. 5. 1.4. Model for statistical pattern recognition . . . . . . . . . . . . . . . . . . .. 7. 1.5. Parallel diagnosis system . . . . . . . . . . . . . . . . . . . . . . . . . . . 13. 1.6. Thesis overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14. 2.1. Flowchart of basic VBM steps . . . . . . . . . . . . . . . . . . . . . . . . 18. 2.2. Concepts of the spatial normalization . . . . . . . . . . . . . . . . . . . . . 19. 2.3. Flowchart of optimized VBM protocol . . . . . . . . . . . . . . . . . . . . 21. 2.4. Illustration of VBM implementation . . . . . . . . . . . . . . . . . . . . . 27. 2.5. BET2 segmentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28. 2.6. Dimension Reduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30. 2.7. Projection Illustration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33. 3.1. Overview of computer-aided system . . . . . . . . . . . . . . . . . . . . . 37. 3.2. Combination of individual classifiers . . . . . . . . . . . . . . . . . . . . . 40. 3.3. A diagram of density estimation . . . . . . . . . . . . . . . . . . . . . . . 44. 3.4. A general form of a ROC curve . . . . . . . . . . . . . . . . . . . . . . . . 49. 4.1. Volumetric atrophy of GM in SCA3 patients . . . . . . . . . . . . . . . . . 55. 4.2. Volumetric atrophy of WM in SCA3 patients . . . . . . . . . . . . . . . . 57. 4.3. Volumetric enlargement of CSF in SCA3 patients . . . . . . . . . . . . . . 58. 4.4. Volumetric atrophy of GM in BD patients . . . . . . . . . . . . . . . . . . 62 v.
(16) 4.5. Volumetric enlargement of WM in BD patients . . . . . . . . . . . . . . . 63. 4.6. Volumetric enlargement of CSF in BD patients . . . . . . . . . . . . . . . 64. 5.1. Predictions of BD patients by varying window size . . . . . . . . . . . . . 88. 5.2. Classification accuray on the BD classifier . . . . . . . . . . . . . . . . . . 89. 5.3. Visualization of density functions in varying the window size . . . . . . . . 90. 5.4. ROC curves of the BD classifier with two PC selection method . . . . . . . 92. 5.5. Cartoon-like representation of classification of three groups . . . . . . . . . 96. vi.
(17) List of Tables 1.1. Comparisons of four CAD systems . . . . . . . . . . . . . . . . . . . . . . 11. 3.1. Interpretations of TP, FP, TN, and FN . . . . . . . . . . . . . . . . . . . . 48. 4.1. Demographic and clinical data of three study groups . . . . . . . . . . . . 52. 4.2. Anatomical interpretation of GM volumetric atrophy in SCA3 patients . . . 56. 4.3. Volumetric atrophy of WM in SCA3 patients . . . . . . . . . . . . . . . . 59. 4.4. Demographic and clinical data of BD study groups . . . . . . . . . . . . . 61. 4.5. Brain structural atrophy in BD patients . . . . . . . . . . . . . . . . . . . . 65. 4.6. Brain structural enlargement in BD patients . . . . . . . . . . . . . . . . . 66. 4.7. Predictions by a SCA3 classifier with variance-based PC selection method . 68. 4.8. Predictions by a SCA3 classifier with variance-based PC selection method . 69. 4.9. Predictions by a SCA3 classifier with variance-based PC selection method . 70. 4.10 Predictions by a SCA3 classifier with significant-based PC selection method 71 4.11 Predictions by a SCA3 classifier with significant-based PC selection method 72 4.12 Predictions by a SCA3 classifier with significant-based PC selection method 73 4.13 Predictions by a BD classifier with variance-based PC selection method . . 75 4.14 Predictions by a BD classifier with variance-based PC selection method . . 76 4.15 Predictions by a BD classifier with variance-based PC selection method . . 77 4.16 Predictions by a BD classifier with significant-based PC selection method . 78 4.17 Predictions by a BD classifier with significant-based PC selection method . 79 4.18 Predictions by a BD classifier with significant-based PC selection method . 80 5.1. Clinical data of experimental groups . . . . . . . . . . . . . . . . . . . . . 84 vii.
(18) 5.2. Performance of GM classifier . . . . . . . . . . . . . . . . . . . . . . . . . 85. 5.3. Performance of WM classifier . . . . . . . . . . . . . . . . . . . . . . . . 86. 5.4. PAUC indices for ROC curves of two BD classifiers. . . . . . . . . . . . . 93. 5.5. Classification of SCA3 patients on the BD classifier . . . . . . . . . . . . . 94. 5.6. Classification of SCA3 patients on the BD classifier . . . . . . . . . . . . . 95. 5.7. Classification of BD patients on the SCA3 classifier . . . . . . . . . . . . . 97. 5.8. Classification of BD patients on the SCA3 classifier . . . . . . . . . . . . . 98. viii.
(19)
相關文件
(4) A principal selection committee shall select in an open, fair and transparent manner a suitable person for recommendation under section 57 from candidates nominated in an open,
Section 3 is devoted to developing proximal point method to solve the monotone second-order cone complementarity problem with a practical approximation criterion based on a new
Abstract In this paper, we consider the smoothing Newton method for solving a type of absolute value equations associated with second order cone (SOCAVE for short), which.. 1
Lin, A smoothing Newton method based on the generalized Fischer-Burmeister function for MCPs, Nonlinear Analysis: Theory, Methods and Applications, 72(2010), 3739-3758..
We have also discussed the quadratic Jacobi–Davidson method combined with a nonequivalence deflation technique for slightly damped gyroscopic systems based on a computation of
For the data sets used in this thesis we find that F-score performs well when the number of features is large, and for small data the two methods using the gradient of the
In the work of Qian and Sejnowski a window of 13 secondary structure predictions is used as input to a fully connected structure-structure network with 40 hidden units.. Thus,
The MTMH problem is divided into three subproblems which are separately solved in the following three stages: (1) find a minimum set of tag SNPs based on pairwise perfect LD
