二維凝膠電泳一直是研究蛋白質上非常重要的一項工具，但是我們必須對凝膠影像作一些處理，才可以從中得到所要的資訊。這些處理過程像是蛋白質點的偵測和比對，若是以人工的方式來做這些分析，需費大量時間外，結果也常常不盡理想。所以我們希望能夠設計一套系統，來幫助相關的研究員能夠方便且迅速的分析結果。例如今天我們有兩張未感染病菌及受感染病菌的二維凝膠電泳影像，因為其中的蛋白質會因為有無感染病菌而有不同的變化，我們先利用之前所偵測出來的結果，來判斷兩張二維凝膠電泳影像上，是不是有哪些蛋白質點產生變化：變大、變小、變濃、變淡或是消失等等。這些蛋白質點就是我們感興趣的，也是作此研究的目的所在，因為這表示了這些蛋白質點在感染了特種病菌之後，會產生反應和變化。因此，我們就根據蛋白質體研究者的需求，設計了這樣的一套系統。系統中，我們主要將先前在二維凝膠電泳影像上所偵測出來的蛋白質點，作後續的比對功能，在這邊我們使用了數學上的方法，先找出幾組對應點當作我們的基準點，再經由這些基準點找出兩張圖上點之間的數學轉換方程式，使其上每個點都能夠滿足這樣的一個方程式；另外我們也根據圖上點的資訊，作些微的調整，以達到所需的比對功能。最後，我們再依據使用者的要求，呈現比對的結果，使得使用者能夠很快的得到他們所要的結果,以便做更進一步的分析。最後，我們將實作後的比對結果顯示出來，說明此系統的比對結果是非常良好的。

In proteomics, 2D gel electrophoresis plays a very important role. We need some processes on these 2D gel electrophoresis images to get information we want. These processes include detection and registration of protein spots. Traditionally, researchers can pick protein spots in the gel images manually. As a result, they spent much time but still made mistakes. For this reason, we proposed a system to assist researchers in dealing with this problem and analyzing protein characteristics. For instance, we got two 2D gel images. One is protein with germs infective, the other is protein with germs anti-infective. In two images, protein spots are different from each other. We take results of detection of protein spots to determine if protein spots change in two images. These changes like getting bigger or smaller, darker or lighter, even disappearing. And, these protein spots are what we are interested. Therefore, we design a system in accordance with demands of researchers. In this system, we mainly take results of detection of protein spots in 2D gel images and develop follow-up capability of matching protein spots. We use methods on mathematics, that is, to select several pairs of spots in two images as landmarks, and then we can find an equation that could transform the source image into the target image. Thus, all spots in images will satisfy this equation and our aim to match these protein spots will be achieved. We show our results of matching depending on demands of users to let them get results efficiently.

目錄
第一章 緒論 1
1.1 研究背景 1
1.2 研究動機 3
1.3 論文組織架構 4
第二章 文獻分析 7
2.1 二維凝膠電泳影像 7
2.2 影像比對的技術 12
第三章 蛋白質點的比對 16
3.1 事前工作 16
3.2 Thin-Plate Splines 18
3.2.1 標記基準點 19
3.2.2 定義函數 20
3.2.3 代數的Thin-Plate Splines 22
3.3 取點修正 29
第四章 系統實作及實驗結果 30
4.1 系統架構 30
4.2 實驗結果 34
4.3 結果分析 51
第五章 結論 59
5.1 結論 59
5.2 未來展望 60
參考文獻 61
附錄 英文論文 65

圖目錄
圖2.1 等電點分離 8
圖2.2 分子量分離 9
圖2.3 二維凝膠電泳影像 10
圖2.4 湯普生轉形方格 14
圖3.1 二維凝膠電泳圖蛋白質點的偵測結果 17
圖3.2  Landmark標記示意圖 20
圖3.3 biharmonic equation的fundamental solution 22
圖4.1 系統畫面 31
圖4.2 檔案 35
圖4.3 執行 36
圖4.4 環境設定 37
圖4.5 開啟來源影像與目的影像 38
圖4.6 設定基準點數為8組 39
圖4.7 選取8組比對基準點，以亮綠色標記 40
圖4.8 經過Thin-Spline Splines技術比對後的初步結果 42
圖4.9 沒有統一中心處理，顯示結果較不精確 44
圖4.10 經過統一中心位置處理，對應較圖4.9良好 45
圖4.11 沒有範圍處理，導致比對不精確 46
圖4.12 經過範圍處理後，結果較圖4.11精確 47
圖4.13 開始進行手動比對 48
圖4.14 手動點選比對 49
圖4.15 手動比對中沒有找到對應點 50
圖4.16 dis500影像與dis1000影像自動比對結果呈現 52
圖4.17 dis500影像與dis1000影像比對結果分析 53
圖4.18 七個基準點比對dis300影像和dis500影像 55
圖4.19 八個基準點比對dis300影像與dis500影像 56
圖4.20 dis200影像與dis300影像比對結果 57

[1]	Bookstein, F.L., “Principal warps: thin-plate splines and the decomposition of deformations,” Pattern Analysis and Machine Intelligence, IEEE Transactions on Volume 11, Issue 6, Page(s):567 – 585, June 1989.
[2]	Khan,N., Rahman, S., “A new approach to detect similar proteins from 2D gel electrophoresis images,” Bioinformatics and Bioengineering, 2003. Proceedings. Third IEEE Symposium on 10-12 Page(s):182 – 189, March 2003.
[3]	Eriksson, A.P., Astrom, K., “ Bijective Image Registration using Thin-Plate Splines,” Pattern Recognition, 2006. ICPR 2006. 18th International Conference on Volume 3, Page(s):798 – 801, 2006.
[4]	Rohr, K., Stiehl, H.S., Sprengel, R., Buzug, T.M., Weese, J., Kuhn, M.H., “ Landmark-based elastic registration using approximating thin-plate splines,” Medical Imaging, IEEE Transactions on Volume 20, Issue 6, Page(s):526 – 534, June 2001.
[5]	Ming-Hung Tsai, “ Detection and Analysis of Protein Spots in 2D Gel Electrophoresis Image,” Master’s Thesis of  Department of Computer Science and Information Engineering, Tamkang University, Taipei, Taiwan, 2006.
[6]	Juin-Lin Kuo, “ Image Analysis System for Protein Two Dimensional Gel Electrophoresis,” Master’s Thesis of  Department of Computer Science and Information Engineering, ChungHua University, Taipei, Taiwan, 2003.
[7]	Hojte, P.B., Xiaoxing Wang, “Registering electrophoresis images for bioinformatics study of protein,” Multimedia and Expo, 2003. ICME '03. Proceedings. 2003 International Conference on Volume 3, 6-9 Page(s): III - 465-8 vol.3, July 2003.
[8]	Frank Hoffmann, Klaus Kriegel, Carola Wenk, “Matching 2D Patterns of Protein Spots,” Proceedings of the fourteenth annual symposium on Computational geometry SCG '98  ACM Press, June 1998.
[9]	Yasein, M.S., Agathoklis, P., “An improved algorithm for image registration using robust feature extraction,” Electrical and Computer Engineering, 2005. Canadian Conference on 1-4 Page(s):1927 – 1930, May 2005.
[10]	Johnson, H.J., Christensen, G.E., “Consistent landmark and intensity-based image registration,” Medical Imaging, IEEE Transactions on Volume 21,Issue 5, Page(s):450 – 461, May 2002.
[11]	Rohr, K., Cathier, P., Worz, S., “Elastic registration of gel electrophoresis images based on landmarks and intensities,” Biomedical Imaging: Macro to Nano, 2004. IEEE International Symposium on 15-18 Page(s):1451 - 1454 Vol. 2, April 2004.
[12]	Science Monthly and King-Taiwan Information Technology Inc., http://203.68.20.65/science/content/1998/00080344/0004.htm
[13]	Wolfram MathWorld, http://mathworld.wolfram.com/