近年來心血管疾病日漸增加，如主動脈栓塞及主動脈瘤等。隨著科技進步也出現了很多檢測方法，像是電腦斷層，核磁共振造影等，主要都是掃描人體的切面影像，提供醫師診斷病情及治療的參考，或是檢查手術後復原情況。由於主動脈疾病在治療上有較大的風險，因此精確的估算血管及腫瘤的面積是非常重要的。一般這種工作需要專業的醫師以人工的方式標示主動脈輪廓，而這通常是一項耗時且費力的工作。
本論文的目的在使用電腦斷層(CT)影像，協助進行半自動的主動脈分割。我們的方法結合了Region Growing和Ray Propagation，使用梯度及亮度資訊，對血管邊界進行分割，並可利用執行結果估計下一張影像的起始位置，做到處理連續影像的效果。我們先透過Region Growing的部分計算出大致的血管輪廓資訊和中心，並使用這些資訊輔助Ray Propagation部份，進行射線數量的估計以及門檻值的計算。Ray Propagation由中心點開始，由射線向外檢查方向梯度及亮度，找尋血管邊界停止。最後透過Region Growing的生長範圍，調整過長或過短的射線。我們的方法對於內部不均勻或是有雜訊血管影像的處理有不錯的效果，亦可處理無注射顯影劑的血管，對輪廓較狹長的血管也能成功的分割。

In this paper we propose a method based on region growing and ray propagation to segment aortic lumen or aortic aneurysm from computed tomography (CT) angiograph images. First, we use an improved region growing method to estimate the roughly range and the centroid of vessel lumen. Then we use a ray propagation method based on gradient and luminance to find the contour of vessel. Finally, we use the information of the first step to refine this contour and obtain the final result. The proposed method does not need to tune any parameter and show a satisfactory result in various CT images. In addition, it can be easily applied to continuous CT images.

目錄
第一章 緒論	1
第二章 相關文獻回顧	7
第三章 研究方法	14
3.1 Region Growing	15
3.2 Ray Propagation	22
3.2.1 Number of Rays	23
3.2.2 Two Thresholds	24
3.2.3 Post Processing	27
第四章 實驗結果與討論	30
第五章 結論與未來展望	42
參考文獻	44
附錄：英文論文	48

圖目錄
圖 1. 胸主動脈及腹主動脈	2
圖 2. 電腦斷層造影儀器	3
圖 3. CT影像中各種組織的HU值分布	4
圖 4. 心導管覆膜支架 [19]	5
圖 5. Region of interest	6
圖 6. Snake [20]	8
圖 7. k-means [21]	9
圖 8. 分水嶺變換 [22]	10
圖 9. 以區域生長做主動脈分割	11
圖 10. 以射線傳遞做主動脈分割 [2]	12
圖 11. Region Growing過程示意圖	16
圖 12. (a)原始CT影像 以及 (b)經過Sobel filter後的梯度資訊	17
圖 13. (a)使用單一seed point無法完整生長之影像；(b)給予起始區域進行生長可以得到比較完整的結果	18
圖 14. Region Growing生長條件	21
圖 15. Region Growing流程圖	22
圖 16. 計算影像之ratio以判斷輪廓形狀	24
圖 17. Ray Propagation流程圖	29
圖 18. 正常含有顯影劑之主動脈	30
圖 19. 不含顯影劑之主動脈	31
圖 20. 胸主動脈瘤及升主動脈	32
圖 21. 主動脈弓	33
圖 22. 動脈瘤及血管支架 有缺口之情形	34
圖 23. 腎臟部分影像	35
圖 24. 比較使用初始點及框選初始範圍之結果	36
圖 25. 分割失敗的範例	37
圖 26. 髂動脈 面積較小的血管	38
圖 27. 動脈瘤以及面積較小的支架	39
圖 28. 形狀較不規則的主動脈影像	40
圖 29. 方法比較	41

[1] 賴景義 教授, 陳信豪,“二維 CT 醫學影像之骨頭輪廓自動擷取”, 國立中央大學 機械工程研究所, 2008.
[2] Hüseyin Tek, Dorin Comaniciu, and James P. Williams, “Vessel Detection by Mean Shift Based Ray Propagation,” pp. 228-235, 2001.
[3] Long Chen, and C. L. Philip Chen, “A Multiple-Kernel Fuzzy C-Means Algorithm for Image Segmentation,” IEEE TRANSACTIONS ON SYSTEMS, MAN, AND CYBERNETICS, pp. 1263-1274, Oct. 2011.
[4] Suthakar Somaskandan, and Sinnathamby Mahesan, ”A Level Set Based Deformable Model for Segmenting Tumors in Medical Images,” Pattern Recognition, Informatics and Medical Engineering, pp. 266-271, Mar. 21-23, 2012.
[5] Hüuseyin Tek, Alper Ayvacı, and Dorin Comaniciu, “Multi-Scale Vessel Boundary Detection”, First International Workshop, CVBIA 2005, pp. 388-398, Oct. 21, 2005.
[6] O. Wink, W.J. Niessen, and M.A. Viergever, "Fast Quantification of Abdominal Aortic Aneurysms from CTA Volumes," Medical Image Computing and Computer-Assisted Interventation- MICCAI’98, pp. 138-145, 1998.
[7] Among Peng, Bernard Dubuisson, and Mohammed Benjelloun, "A Study on the Forms of Smoothing Filters for Step and Ramp Edge Detection," Pattern Recognition, 1992. Vol.III. Conference C: Image, Speech and Signal Analysis, Proceedings., 11th IAPR International Conference on, pp. 741-744, Aug.30-Sep. 3 1992.
[8] Chung-Yi Huang, Lai-Jun Luo, Wen-Teng Wang, Pei-Yuan Lee, Shang-Chih Lin, and Jiing-Yih Lai, "Efficient Segmentation Algorithm for 3D Bone Models Construction on Medical Images," Journal of Medical and Biological Engineering, pp. 375-386, Dec. 1 2010.
[9] David Lesage, Elsa D. Angelini, Isabelle Bloch, and Gareth Funka-Lea, "A review of 3D vessel lumen segmentation techniques: Models, features and extraction schemes," Includes Special Section on Computational Biomechanics for Medicine, Volume 13, Issue 6, pp. 819-940, Dec. 2009.
[10] Michael Gayhart, and Keisuke Yoshida, "Automated Segmentation of the Aortic Artery: Evaluation of Images Obtained by Triple Rule-Out Protocol," Image and Signal Processing (CISP), 2011 4th International Congress on, Volume 4, pp. 1833-1836, 15-17 Oct. 2011.
[11] Gabriela Castellano, and Roberto A. Lotufo, "Sulcal Lines Extraction for Cortex Characterization in Cerebral MR Images using the Image Foresting Transform," Computer Graphics and Image Processing, 2003. SIBGRAPI 2003. XVI Brazilian Symposium on, pp. 355-362, 12-15 Oct. 2003.
[12] Rocío Cabrera Lozoya Lozoya, Olivier Bouchot, Tadeusz Sliwa, Eric Steinmetz, and Yvon Voisin, Alain Lalande, "A 4D patient-specific modelling of the thoracic aorta from cine-MR Images," Signal-Image Technology and Internet-Based Systems (SITIS), 2011 Seventh International Conference on, pp. 269-276 Nov. 28 2011-Dec. 1 2011.
[13] C.W. Chen, J. Luo, K.J. Parker, "Image segmentation via adaptive K-mean clustering and knowledge based morphological operations with biomedical applications", IEEE Transactions on Image Processing, Vol.7 (12), pp.1673-1683, 1998.
[14] R. Hummel, "Image Enhancement by Histogram Transformation", Computer Graphics and Image Understanding, vol. 6, pp.184-195 1977.
[15] R. Haralick, "Digital step edges from zero crossing of second directional derivatives", IEEE Trans. on Pattern Analysis and Machine Intelligence, 6(1):58-68, 1984.
[16] S. W. Zucker "Region growing: Childhood and adolescence", Comput. Graph. Image Process. , vol. 5, pp.382-399, 1976.
[17] M. Kass, A. Witkin, and D. Terzopoulos, "Snakes: Active contour models," International Journal of Computer Vision, vol. V1, no. 4, pp. 321-331, January 1988.
[18] S. Beucher and C. Lantuejoul , "Use of watersheds in contour detection", Proc. Int. Workshop Image Processing, Real-Time Edge and Motion Detection I Estimation, 1979.
[19] http://www.tacacare.org/instruction.php?instruction_id=3
[20] http://imagejdocu.tudor.lu/doku.php?id=plugin:segmentation:active_contour:start
[21] http://informationandvisualization.de/blog/kmeans-and-voronoi-tesselation-built-processing
[22] http://en.wikipedia.org/wiki/Watershed_%28algorithm%29