超音波照影是初期肝硬化的診斷工具之一，可以藉由比較同一病人肝與脾的超音波影像紋理來判定此病人是否患有肝硬化。本篇論文研究新的統計比較分析方法，應用在台大醫院內科視鏡超音室收集來的臨床超音波影像資料上。我們結合不同的影像特徵選取方法、降維度工具及統計分類規則，然後運用Jackknife和Bootstrap的方法來估計影像分類誤差以找尋最佳的肝硬化超音波影像分析方法。

Ultrasound images are one of the main diagnosis tools for early detection and regular check ups for liver cirrhosis. By comparing the echo-textures of liver and the accompanied spleen of the same person at the same scan, the variations between subjects, systems and machines are under control. This study proposes new statistical comparison methods for this purpose based on the clinic data collected at the National Taiwan University Hospital. The Jackknife and bootstrap prediction errors for different methods are investigated. Different combinations of feature selections, dimension reduction techniques and classification rules are studied and reported.

1 Introduction.................................................1
2 Diagnosis Criteria of Liver Cirrhosis in Ultrasonic Images...3
2.1 Medical Criteria........................................3
2.2 Computer Assisted Comparisons of Echo-Textures of Livers and Spleens....................................................5
3 Feature Selections of Ultrasonic image.......................6
3.1 Local Blocks in the Space Domains........................6
3.2 Local Blocks in the Frequency Domain.....................7
3.3 Local Blocks in the Space-Frequency Domain...............9
4 Dimension Reduction Techniques..............................14
4.1 Principal Component Analysis (PCA)......................14
4.2 Sliced Inverse Regression (SIR).........................15
5 Construction Methods of Classification Rules................17
5.1 Model Checking..........................................17
5.2 Discriminant Analysis...................................17
6 Empirical Studies Based on Simple Heuristic Methods.........20
6.1 Individual Statistics of Liver and Spleen Images........20
6.2 Pairwise Comparisons of Liver and Spleen Images.........26
7 Classification with Dimension Reduction Techniques..........31
7.1 Projection onto the PCA Directions of Livers and Spleens31
7.2 Projection onto the PCA Directions of Spleens...........39
7.3 Projection onto the SIR Directions of Livers and Spleens47
8 Jackknife and Bootstrap Estimates of Error Rates............51
8.1 Jackknife Estimates.....................................51
8.2 Bootstrap Estimates.....................................51
8.3 Empiric Results.........................................52
9 Conclusion and Discussion74

[1] T. W. Anderson, An Introduction to Multivariate Statistical Analysis, Wiley, 1984.
[2] L. Breiman, J. H. Friedman, R. A. Olshen, and J. C. Stone, Classification and Regression Trees, Wadsworth, CA: Belmant, 1984.
[3] C.-H. Chen and K.-C. Li, ``Discriminant analysis, tree-structured classification, and sliced inverse regression,'' Technical Report, Institute of Statistical Science, Academia Sinica, Taipei, Taiwan, 1994.
[4] V. Cherkassky and F. Mulier, Learning from Data: Concepts, Theory, and Methods, Wiley, 1998.
[5] B. Efron, “Estimating the error rate of a prediction rule: improvements of croos-validation,” J. Amer. Statist. Assoc. 1983; 78: 316-331.
[6] B. Efron and R. Tibshirani, An Introduction to the Bootstrap, Chapman and Hall, New York, 1993.
[7] D. J. Hand, Construction and Assessment of classification Rules, Wiley, 1997.
[8] R. C. Gonzalez and R. E. Woods, Digital Image Processing, Addison Wesley, Reading, Mass., 1992.
[9] A. K. Jain and F. Farrokhnia, ``Unsupervised texture segmentation using Gabor filters,'' Pattern Recognition, 24, 12, 1167─1186, 1991.
[10] H. H.-S. Lu, C.-M. Chen, and Y.-C. Shen, “Toward Texture Segmentation Using Wavelets Based on a Vision Model,” Technical Report, National Chiao Tung University, Hsinchu, Taiwan, 1997.
[11] H. H.-S. Lu and D.-I. Yan, ''Texture Segmentation by Iterated Slice Inverse Regression and Space-Frequency Analysis,'' Technical Report, National Chiao Tung University, Hsinchu, Taiwan, 1998.
[12] K. C. Li, “Sliced inverse regression for dimension reduction, (with discussions),'' J. Amer. Statist. Assoc. 1991; 86: 316-342.
[13] C. B. Lin, C. M. Chen and S. Su, “A Computational Early Vision Model for Segmentation of Clinical Ultrasound Images,” 19th Annual International Conference of the IEEE EMBS, Chicago, Illinois, U.S.A., Oct. 1997.
[14] J. Malik and P. Perona, ``Preattentive texture discrimination with early vision mechanisms,'' J. Opt. Soc. Am., A, 7, 923─932, 1990.
[15] K. V. Mardia, Multivariate Analysis, Academic Press, 1979.
[16] G. Nakhaeizadeh and C. C. Taylor, Machine Learning and Statistics: the Interface, Wiley, 1997.
[17] R. H. Randles and D. A. Wolfe, Introduction to the Theory of Nonparametric Statistics, Wiley, 1979.
[18] T. R. C. Read and N. A. C. Cressie, Goodness-of-Fit Statistics for Discrete Multivariate Data, Springer, New York, 1988.
[19] D. W. Scott, Multivariate Density Estimation, Wiley, New York, 1992.
[20] B. W. Silverman, Density Estimation for Statistics and Data Analysis, Chapman and Hall, London, 1986.
[21] 竹原靖明 等原著, 趙世晃譯, 腹部超音波速成 (腹部ECHO ABC), 合記, 臺北市, 1990.