果蠅腦中嗅覺小球影像的三維切割研究｜國立清華大學博碩士論文庫

簡易檢索 / 詳目顯示

回結果列表

研究生：	王義明 Wang, Yi-Ming
論文名稱：	果蠅腦中嗅覺小球影像的三維切割研究 3D Segmentation for Glomeruli of Drosophila Brain Image
指導教授：	陳永昌 Chen, Yung-Chang
口試委員:	黃文良盧鴻興
學位類別：	碩士 Master
系所名稱：	電機資訊學院 - 電機工程學系 Department of Electrical Engineering
論文出版年：	2012
畢業學年度：	100
語文別：	英文
論文頁數：	50
中文關鍵詞：	果蠅腦、嗅覺小球、切割、分水嶺演算法
外文關鍵詞：	Watershed algorithm, Gradient Vector Flow
相關次數：	點閱：2 下載：0
分享至:	分享至facebook 分享至twitter

查詢本校圖書館目錄查詢臺灣博碩士論文知識加值系統勘誤回報

近年來科學家一直致力於人類腦部的研究，但是由於人腦過於複雜，而且可用來實驗的樣本也相較之下不易取得，所以科學家選擇了功能上有許多相似，但是結構上明顯簡單許多的果蠅腦來當作實驗的題材。由於嗅覺系統控制著果蠅複雜的行為，科學家希望從嗅覺系統中較原始的器官－Antenna lobe開始研究。

為了了解果蠅體內不同基因對嗅覺造成的影響，科學家在不同的果蠅身上抑制不同基因的表現，再藉由其行為反應來推測不同基因的作用。在分析的過程中，需要一個果蠅嗅小球（嗅小球位於Antenna lobe中）的模型，可以方便辨識在不同的基因控制之下，對應的嗅小球和其相關神經的分布為何。至於如何製作一個果蠅嗅小球的模型，就需要從不同的果蠅腦中，切割出當中的嗅小球，再製作出一個嗅小球的模型。

這個研究採用的切割技術是Gradient Vector Flow(GVF) Snake，其捕捉範圍(Capture range)大，和對於初始輪廓(Initial contour)與實際邊界相近的要求不高。在這樣的特性之下，讓我們不需人工輔助，大概提供一個圓形的輪廓去給GVF變形，就可以得到不錯的結果。至於這個初始輪廓到底要設在影像中的哪個位置，這邊我們提出一個嶄新的想法：我們將影像經過一些前處理後，透過分水嶺演算法(Watershed algorithm)粗略地將嗅小球分開。再取每個區域的質心當成是初始圓形輪廓的圓心，如此來決定初始輪廓的位置。

透過我們提出的演算法，不需人工輔助，便可以提供一些切割的結果，最後再交由專家人工判斷哪些結果是可以被接受，可用於建構嗅小球模型或其他後續研究。從實驗結果中可以看出這個演算法提供出的切割結果可有相當程度的貢獻。

In recent years, scientists have been dedicated to the research on human brain. However, the structure of human brain is too complicated, leading scientists in the search for a subject with similar functionalities but simpler structure. Therefore fruit flies, Drosophila melanogaster, turned out to be a good choice. Researchers tend to start with a primal organ in the olfactory system, the antenna lobe (AL), by observing the complex behaviors controlled by the system.

In order to understand the effects of different odorant receptor genes, scientists control different genes and observe the change in behaviors. To analyze the data, a standard glomeruli model (glomeruli are in the AL) would be required to identify which glomeruli and neoropils are controlled by which genes. As for how to generate a glomeruli model, delineations of glomeruli are needed in order to construct the model.

The segmentation method used in our work is Gradient Vector Flow (GVF) Snake. We favor its property of large capture range and less-dependency of initialization. Under these properties, we would be able to obtain a satisfying result simply by providing a circular initial contour. As for where to locate the initial contour, we’ve proposed a whole new method: we separate the glomeruli roughly by watershed algorithm. Then we calculate the center of mass for each region and use them as the center of circle for each initial contour.

We are able to propose contours without manual assistance. Validation of the resultant contours would be later on evaluated by experts. From the experimental results, we may expect this algorithm to be quite contributive.

Table of Contents
摘要    i
Abstract    ii
Table of Contents    iii
List of Figures    v
List of Tables    vii

Chapter 1 Introduction    1
1.1 Background    1
1.2 Motivation    4
1.3 Related Work    4
1.4 Thesis Organization    5
Chapter 2 Research Framework    6
Chapter 3 Seed Determination    9
3.1 Preprocessing    10
3.2 Watershed Algorithm    14
3.3 Valid-seed Check    17
3.4 Summary    22
Chapter 4 Gradient Vector Flow    24
4.1 Preprocessing    25
4.1.1 Watershed Algorithm    25
4.1.2 Gaussian Blur    25
4.1.3 Morphological Reconstruction    27
4.1.4 Sobel Filtering    27
4.1.5 Thresholding    28
4.2 Gradient Vector Flow    30
4.3 Post-processing    32
4.3.1 Contour Smoothing    33
4.3.2 Contour for Next Iteration    33
4.4 Stopping Condition    34
4.5 Manual Evaluation    35
4.6 Summary    36
Chapter 5 Experimental Results and Discussions    37
5.1 Results and Evaluation for the Stack with Ground Truth    37
5.1.1 Segmentation Results of a Glomerulus    37
5.1.2 Resultant Contours of Glomeruli on a Single Slice    40
5.1.3 Evaluation of Segmentation Result    42
5.2 Resultant Contours of Other Image Stacks    42
5.3 Discussion    43
Chapter 6 Conclusions and Future Work    46
6.1 Conclusions    46
6.2 Future Work    47
Reference    49

                                

[1]H. H. Lin, J. S. Y. Lai, A. L. Chin, Y. C. Chen and A. S. Chiang, "A Map of Olfactory Representation in the Drosophila Mushroom Body," Cell, no. 128, pp. 1205-1217, 2007.
[2]A. Couto, M. Alenius and B. J. Dickson, "Molecular, Anatomical and Functional Organization of the Drosophila Olfactory System," Current Biology, no. 15, pp. 1535-1547, 2005.
[3]L. B. Vosshall and A. M. Wong, "An Olfactory Sensory Map in the Fly Brain," Cell, vol. Vol. 102, pp. 147-159, 2000.
[4]C. H. Lai, "3D Registration and Segmentation for Glomeruli of Drosophila Brain," Master Thesis, National Tsing Hua University, 2010.
[5]J. Y. Guo, "Glomeruli of the Drosophila Brain: 2D Image Analogy and Semi-automatic Edge Detection," Master Thesis, National Tsing Hua University, 2010.
[6]P. P. Laissue and C. Reiter, "Three-Dimensional Reconstruction of the Antennal Lobe in Drosophila melanogaster," The Journal of Comparative Neurology, no. 405, pp. 543-552, 1999.
[7]"Nikon MicroscopyU|Confocal Microscopy|Basic Concepts," [Online]. Available: http://www.microscopyu.com/articles/confocal/confocalintrobasics.html.
[8]C. Xu and J. L. Prince, "Snakes, Shapes, and Gradient Vector Flow," IEEE Transactions on Image Processing, vol. 7, no. 3, pp. 359-369 , 1998.
[9]F. Meyer and S. Beucher, "Morphological Segmentation," J. Visual Comm. and Image Representation, vol. vol. 1, no. no. 1, pp. 21-46, 1990.
[10]M. Kass, A. Witkin and D. Terzopoulos, "Snake: Active contour models," Int. J. Comput. Vis., vol. vol. 1, pp. 321-331, 1987.
[11]L. R. Dice, "Measures of the Amount of Ecologic Association Between Species," Ecology, vol. 26, no. 3, p. 297–302, 1945.
[12]Y. C. Chen, "Framework for Creation of the Drosophila Standard Brain from Confocal Microscopy Images," Ph. D. Thesis, National Tsing Hua University, 2007.
[13]G. Lin , U. Adiga, K. Olson, J. F. Guzowski, C. A. Barnes and B. Roysam, "A Hybrid 3D Watershed Algorithm Incorporating Gradient Cues and Object Models for Automatic Segmentation of Nuclei in Confocal Image Stacks," Cytometry Part A, vol. vol. 56A, no. 1, pp. 23-36, 2003.
[14]E. Hodneland, N. V. Bukoreshtliev, T. W. Eichler, X. C. Tai, S. Gurke , A. Lundervold and H.-H. Gerdes, "A Unified Framework for Automated 3-D Segmentation of Surface-Stained Living Cells and a Comprehensive Segmentation Evaluation," IEEE Transactions on Medical Imaging, vol. vol.28, no. 5, pp. 720-737, 2009.
[15]N. Harder, M. Bodnar, R. Eils, D. L. Spector and K. Rohr, "3D Segmentation and Quantification of Mouse Embryonic Stem Cells in Fluorescence Microscpy Images," 2011 IEEE International Symposium on Biomedical Imaging: From Nano to Macro, pp. 216 - 219, 2011.
[16]R. C. Gonzalez and R. E. Woods, Digital Image Processing, 3rd ed. ed., Prentice Hall, 2007.

全文公開日期本全文未授權公開 (校內網路)
全文公開日期本全文未授權公開 (校外網路)

簡易檢索 / 詳目顯示

相關論文