臺灣博碩士論文加值系統

乳癌是20-59歲的女性死亡率最高的癌症，及早的治療，可有效降低乳癌的致死率。近年來，乳房超音波影像常用來診斷腫瘤的良惡性，為了提升正確率，大部分研究在腫瘤分類前先做腫瘤切割，切割的結果直接影響到了良惡性腫瘤的判斷。因此，此篇研究主要的目的是使用image matting改善腫瘤的切割結果，進行腫瘤的電腦輔助診斷。首先，從超音波影像取出腫瘤的VOI (volume-of-interest)區域，並透過一連串的影像處理方法進行初步切割，而後將預切割的結果建立成tri-map，接著根據tri-map所畫出的未知區域使用image matting進行腫瘤切割，從切割結果擷取出紋理特徵和形態特徵，最後，利用SVM分類器以及擷取出的特徵進行腫瘤良惡性的分類。在此次研究中採用了80個病例，包含40個良性腫瘤以及40個惡性腫瘤，研究結果顯示，使用image matting確實比不使用image matting有更佳的切割效能，並且在特徵上可得到GLCM (gray-level co-occurrence matrix)、ranklet以及橢圓擬合特徵的組合較具有分辨率，可達到準確率85.0% (68/80)、靈敏性87.5% (35/40)、特異性 82.5% (33/40)，以及ROC曲線面積0.8829。從結果上來看，使用image matting確實能夠改善腫瘤的切割並且得到較為準確的良惡性分類結果。

In women aged 20-59 years, breast cancer is the highest mortality. Early treatment can effectively reduce the mortality of breast cancer. In recently, breast ultrasound image is often used to diagnose between benign and malignant tumors. For increasing the accuracy, most researches segment the tumor before classification, and the segmented results directly affect the classification between benign and malignant tumors. Therefore, the purpose of this study is to refine segmented tumors using the image matting method for computer-aided diagnosis. First, the volume-of-interest (VOI) of tumor was extracted from the ultrasound image and pre-segmented by a conventional segmentation method. The tri-map including the background, foreground, and unknown region was created with the pre-segmented tumor, and then the image matting method was applied for refining the segmentation according to the unknown region of tri-map. Texture and morphology features were extracted from refined segmentation result and then the support vector machine was applied with extracted features to classify tumor into benign or malignant tumor. This study was validated with 80 cases including 40 benign and 40 malignant breast lesions. According to the experiment results, applying the image matting method had better performance than not applying the image matting method, and the combination of GLCM, ranklet, and ellipsoid fitting feature set had significant (resolution??. The accuracy, sensitivity, specificity, and the area under ROC achieved 85.0% (68/80), 87.5% (35/40), 82.5% (33/40), and 0.8829, respectively. From the experiment results, the image matting method could actually refine tumor segmentation, and more precise classification between benign and malignant tumor results were obtained.

口試委員會審定書.........................................ii
致謝..................................................iii
摘要...................................................iv
Abstract...............................................v
List of Figures......................................vii
List of Tables......................................viii
Chapter 1 Introduction.................................1
Chapter 2 Material.....................................5
2.1. Patients and Lesion Characters.................5
2.2. Data Acquisition...............................5
Chapter 3 The Tumor Diagnosis System...................7
3.1. VOI extraction.................................8
3.2. Tumor Segmentation.............................9
3.2.1. Tri-map creation..............................10
3.2.2. Sigmoid Filter................................10
3.2.3. Recursive Gaussian Smoothing Filter...........11
3.2.4. Connected Component Filter and Label Mapping..11
3.2.5. Region Construction...........................12
3.3. 3-D Image Matting.............................13
3.4. Feature Extraction............................15
3.4.1. Texture features..............................15
3.4.2. Morphological features........................17
3.5. Classification................................22
3.5.1. Tumor Classification..........................22
Chapter 4 Experiment Results and Discussion...........24
4.1. Experiment Environment........................24
4.2. Feature Analysis..............................24
4.3. Results.......................................26
4.4. Discussion....................................33
Chapter 5 Conclusion and Future Work..................36
References............................................37

[1]R. L. Siegel, K. D. Miller, and A. Jemal, "Cancer Statistics, 2016," Ca-a Cancer Journal for Clinicians, vol. 66, pp. 7-30, Jan-Feb 2016.
[2]A. Takemura, A. Shimizu, and K. Hamamoto, "Discrimination of breast tumors in ultrasonic images using an ensemble classifier based on the AdaBoost algorithm with feature selection," IEEE Trans Med Imaging, vol. 29, pp. 598-609, Mar 2010.
[3]A. Takemura, A. Shimizu, and K. Hamamoto, "Discrimination of Breast Tumors in Ultrasonic Images Using an Ensemble Classifier Based on the AdaBoost Algorithm With Feature Selection," Ieee Transactions on Medical Imaging, vol. 29, pp. 598-609, Mar 2010.
[4]A. S. Hong, E. L. Rosen, M. S. Soo, and J. A. Baker, "BI-RADS for sonography: Positive and negative predictive values of sonographic features," American Journal of Roentgenology, vol. 184, pp. 1260-1265, Apr 2005.
[5]D. R. Chen, R. F. Chang, W. J. Kuo, M. C. Chen, and Y. L. Huang, "Diagnosis of breast tumors with sonographic texture analysis using wavelet transform and neural networks," Ultrasound in Medicine and Biology, vol. 28, pp. 1301-1310, Oct 2002.
[6]W. C. Shen, R. F. Chang, W. K. Moon, Y. H. Chou, and C. S. Huang, "Breast ultrasound computer-aided diagnosis using BI-RADS features," Academic Radiology, vol. 14, pp. 928-939, Aug 2007.
[7]U. R. Acharya, M. R. K. Mookiah, S. V. Sree, D. Afonso, J. Sanches, S. Shafique, et al., "Atherosclerotic plaque tissue characterization in 2D ultrasound longitudinal carotid scans for automated classification: a paradigm for stroke risk assessment," Medical & Biological Engineering & Computing, vol. 51, pp. 513-523, May 2013.
[8]L. Y. Cai, X. Wang, Y. Y. Wang, Y. Guo, J. H. Yu, and Y. Wang, "Robust phase-based texture descriptor for classification of breast ultrasound images," Biomedical Engineering Online, vol. 14, Mar 24 2015.
[9]M. S. Kavic, "Three-dimensional ultrasound," Surgical Endoscopy-Ultrasound and Interventional Techniques, vol. 10, pp. 74-76, Jan 1996.
[10]D. B. Downey, A. Fenster, and J. C. Williams, "Clinical utility of three-dimensional US," Radiographics, vol. 20, pp. 559-71, Mar-Apr 2000.
[11]K. M. Kelly, J. Dean, W. S. Comulada, and S. J. Lee, "Breast cancer detection using automated whole breast ultrasound and mammography in radiographically dense breasts," European Radiology, vol. 20, pp. 734-742, Mar 2010.
[12]E. Wenkel, M. Heckmann, M. Heinrich, S. A. Schwab, M. Uder, R. Schulz-Wendtland, et al., "Automated breast ultrasound: Lesion detection and BI-RADS (TM) classification - a pilot study," Rofo-Fortschritte Auf Dem Gebiet Der Rontgenstrahlen Und Der Bildgebenden Verfahren, vol. 180, pp. 804-808, Sep 2008.
[13]Y. Chou, C. Tiu, H. Chiang, S. Chen, H. Chiou, and S. Chiou, "Ultrasound ACR BI-RADSR Categories Applied in an Automated Breast Ultrasound System: Diagnostic Reliability," in Radiological Society of North America scientific assembly and annual meeting program. Oak Brook, Ill: Radiological Society of North America, 2006.
[14]W. K. Moon, Y. W. Shen, C. S. Huang, L. R. Chiang, and R. F. Chang, "Computer-Aided Diagnosis for the Classification of Breast Masses in Automated Whole Breast Ultrasound Images," Ultrasound in Medicine and Biology, vol. 37, pp. 539-548, Apr 2011.
[15]J. H. Kim, J. H. Cha, N. Kim, Y. Chang, M. S. Ko, Y. W. Choi, et al., "Computer-aided detection system for masses in automated whole breast ultrasonography: development and evaluation of the effectiveness," Ultrasonography, vol. 33, pp. 105-15, Apr 2014.
[16]S. C. Chang, Y. C. Lai, Y. H. Chou, and R. F. Chang, "Breast elastography diagnosis based on dynamic sequence features," Medical Physics, vol. 40, Feb 2013.
[17]R.-F. Chang and C.-M. Lo, "Computer-aided diagnosis for B-mode, elastography and automated breast ultrasound," in International Workshop on Digital Mammography, 2014, pp. 9-15.
[18]A. Levin, D. Lischinski, and Y. Weiss, "A closed-form solution to natural image matting," Ieee Transactions on Pattern Analysis and Machine Intelligence, vol. 30, pp. 228-242, Feb 2008.
[19]H. Rajaguru and V. K. Bojan, "Performance Analysis of EM, SVD, and SVM Classifiers in Classification of Carcinogenic Regions of Medical Images," International Journal of Imaging Systems and Technology, vol. 24, pp. 16-22, Mar 2014.
[20]M. H. Hyun, S. Y. Kim, Y. S. Kang, and Y. S. Ho, "Multiview Foreground Extraction and Composition to Multiview Background Using Trimap Sharing for Natural 3D Scene Generation," International Journal of Imaging Systems and Technology, vol. 20, pp. 285-293, 2010.
[21]O. Juan and R. Keriven, "Trimap segmentation for fast and user-friendly alpha matting," Variational, Geometric, and Level Set Methods in Computer Vision, Proceedings, vol. 3752, pp. 186-197, 2005.
[22]G. Farneback and C. F. Westin, "Improving Deriche-style recursive Gaussian filters," Journal of Mathematical Imaging and Vision, vol. 26, pp. 293-299, Dec 2006.
[23]Q. Dong, "Largest Connected Component of a k-ary n-cube with Faulty Vertices," Journal of Information Science and Engineering, vol. 29, pp. 777-783, Jul 2013.
[24]J. S. Suri, Advances in diagnostic and therapeutic ultrasound imaging. Boston ; London: Artech House, 2008.
[25]R. L. Carceroni, "Journal of mathematical imaging and vision: SIBGRAPI 2006 special issue," Journal of Mathematical Imaging and Vision, vol. 29, pp. 93-94, Nov 2007.
[26]M. Singh, S. Singh, and S. Gupta, "An information fusion based method for liver classification using texture analysis of ultrasound images," Information Fusion, vol. 19, pp. 91-96, Sep 2014.
[27]W. J. Wu and W. K. Moon, "Ultrasound breast tumor image computer-aided diagnosis with texture and morphological features," Academic Radiology, vol. 15, pp. 873-880, Jul 2008.
[28]R. M. Haralick, Shanmuga.K, and I. Dinstein, "Textural Features for Image Classification," Ieee Transactions on Systems Man and Cybernetics, vol. Smc3, pp. 610-621, 1973.
[29]V. K. Sudarshan, E. Y. K. Ng, U. R. Acharya, S. M. Chou, R. S. Tan, and D. N. Ghista, "Computer-aided diagnosis of Myocardial Infarction using ultrasound images with DWT, GLCM and HOS methods: A comparative study," Computers in Biology and Medicine, vol. 62, pp. 86-93, Jul 1 2015.
[30]W.-S. Chen, R.-H. Huang, and L. Hsieh, "Iris recognition using 3D co-occurrence matrix," in International Conference on Biometrics, 2009, pp. 1122-1131.
[31]M. C. Yang, W. K. Moon, Y. C. F. Wang, M. S. Bae, C. S. Huang, J. H. Chen, et al., "Robust Texture Analysis Using Multi-Resolution Gray-Scale Invariant Features for Breast Sonographic Tumor Diagnosis," Ieee Transactions on Medical Imaging, vol. 32, pp. 2262-2273, Dec 2013.
[32]E. Bribiesca, "An easy measure of compactness for 2D and 3D shapes," Pattern Recognition, vol. 41, pp. 543-554, Feb 2008.
[33]L. A. Meinel, A. H. Stolpen, K. S. Berbaum, L. L. Fajardo, and J. M. Reinhardt, "Breast MRI lesion classification: improved performance of human readers with a backpropagation neural network computer-aided diagnosis (CAD) system," J Magn Reson Imaging, vol. 25, pp. 89-95, Jan 2007.
[34]K. F. Mulchrone and K. R. Choudhury, "Fitting an ellipse to an arbitrary shape: implications for strain analysis," Journal of Structural Geology, vol. 26, pp. 143-153, 2004.
[35]L. G. Daniel, "Basic Statistical-Analysis, 3rd Edition - Sprinthall,Rc," Educational and Psychological Measurement, vol. 50, pp. 710-716, Fal 1990.
[36]M. G. Shashaty, J. P. Reilly, C. A. Sims, D. N. Holena, D. Qing, C. M. Forker, et al., "Plasma Levels of Receptor Interacting Protein Kinase-3 (RIP3), an Essential Mediator of Necroptosis, are Associated with Acute Kidney Injury in Critically Ill Trauma Patients," Shock, vol. 46, pp. 139-43, Aug 2016.
[37]H. Ma, A. I. Bandos, and D. Gur, "On the use of partial area under the ROC curve for comparison of two diagnostic tests," Biometrical Journal, vol. 57, pp. 304-320, Mar 2015.
[38]C. C. Meltzer, P. E. Kinahan, P. J. Greer, T. E. Nichols, C. Comtat, M. N. Cantwell, et al., "Comparative evaluation of MR-based partial-volume correction schemes for PET," Journal of Nuclear Medicine, vol. 40, pp. 2053-2065, Dec 1999.