臺灣博碩士論文加值系統

臨床體外試驗抗生素藥敏性方式有非常多種，本實驗以臨床檢體進行抗生素藥敏性快速測試平台，藉由介電泳晶片用於抗生素藥敏性快速測試技術開發所得圖像，進行自動化判讀。在實驗中，提出了一個細菌圖像切割以及判別細菌抗藥性的方法，基於介電泳影像分析的快速檢測平台系統，可應用於臨床檢測細菌的藥物抗藥性試驗。首先，觀察晶片的特性將電極線的部分使用Sobel Detection方法將影像轉成二值化影像，並在水平與垂直方向進行投影值分析等工作。取出對應點與線段，進而求出影像中四塊主要細菌會附著的地帶，這將是偵測細菌用的有興趣區域(Region of interest, ROI)。接著，在ROI區域執行去除雜訊和局部二值化影像以及篩選細菌，以取得細菌所在區域，作為進一步分割的影像資料。再來在將偵測過後所有候選細菌的位置與特徵資訊，透過分析每隻候選細菌的骨架方式計算出細菌長，並使用DP(Dynamic Programming)去優化ACM(Active contour model)演算法取得最佳輪廓解，作為進一步分析的數據資料。由於對藥物敏感的細菌會有延長(elongation)或消失(lysis)的現象，所以本實驗將藉由細菌長度的資訊來判斷該細菌是否具有抗藥性，最後使用細菌的分割結果進行長度的計算去做判定，本實驗所開發的系統會將判讀結果分為兩類，分別為具抗藥性(Resistance)、不具抗藥性(Sensitive)。經由本系統快速有效之影像分析，可以建立抗生素藥劑對於細菌的關聯性，提供醫師臨床診斷與治療細菌感染之重要參考依據。此檢測平台可以提高藥敏性試驗的精確度和效率。且它不僅縮短了傳統檢測方法的處理時間，也節省了寶貴的臨床實驗的成本。

There are many methods for antibiotic susceptibility test (AST) in clinical applications. In this thesis, a fast examination platform is developed for this purpose. Dielectrophoresis image is acquired for automatic discrimination. A rapid evaluation platform based on DEP image analysis has been proposed for antimicrobial susceptibility testing in clinical applications. First, the characteristics of the wafer were observed. The Sobel detection method was used to convert the image of electrode line into a binary image. Then, the projection analysis was performed in both the horizontal and vertical directions. After finding the corresponding points and line segments, the four major zones where the bacteria usually attached were defined as the region of interest, ROI-. Next, the noise reduction and local binarization were performed on the ROI regions, the regions of the bacteria were then obtained for the subsequent bacteria segmentation. The bacterial length of each candidate bacterium was then calculated by using the skeletal pattern. The Dynamic Programming (DP) was then employed to optimize the ACM (Active contour model) and obtained the best profile solution as the refinement. Due to the elongation or lysis phenomenon of drug-sensitive bacteria, the bacteria which was sensitive to the applied antibiotic could be discriminated. Our system classified the interpretation into two categories- the resistance group and the sensitive group. The given bacteria image could be analyzed rapidly and effectively. The proposed system can successfully establish a connection between the antibiotics and bacteria; it also provides an important reference for the physician in clinical diagnosis and treatment of bacterial infection. The accuracy and efficiency for antimicrobial susceptibility testing is improved. It does not only shorten the long processing time required by the traditional testing but save the valuable labor cost in clinical applications.

CONTENTS
摘要 I
Abstract II
Acknowledgements IV
List of Tables VIII
List of Figures IX
Chapter 1 Introduction 1
1.1 Motivation 1
1.2 Related Work 2
1.3 Overview of the Proposed Methods and Thesis Organization 3
Chapter 2 Experimental Materials and Image Preprocessing 5
2.1 Experimental Materials 5
2.1.1 Image data collection and renaming 5
2.1.2 Dielectrophoresis image and bacterial types 6
2.2 Image Preprocessing 7
2.2.1 Resize image 7
2.2.2 The ROI position and the specific point of the processed image 8
2.2.3 Processing electrode 14
Chapter 3 Bacterial Detection 19
3.1 Median filter noise reduction 20
3.2 The Sobel image processing 21
3.3 ROI processing 22
3.3.1 Adjust the ROI 22
3.3.2 Region-of-interest (ROI) extraction 23
3.4 Local binarization 24
3.5 Bacteria filtering 25
3.5.1 Obtain the image object information 25
3.5.2 Filtering conditions 26
3.6 Bacteria elongation 26
3.7 Obtain the boarder position box of each bacterium 29
3.7.1 Connected-component labeling 29
3.7.2 Combine the overlapping boxes 30
Chapter 4 Bacterial Segmentation 31
4.1 Bacterial Skeleton 32
4.1.1 Image preprocessing 32
4.1.2 Trough point 33
4.1.3 Classification 35
4.1.4 Remove noise and move average to get bacterial skeleton 36
4.1.5. Judgment and bacterial length assessment 37
4.2 Bacterial Contour 38
4.2.1 Gradient image processing 39
4.2.2 Initial contour 40
4.2.3 Optimal contour 41
Chapter 5 Drug Resistance Analysis 44
5.1 Standard for drug resistance judgment 44
Chapter 6 Experimental Results and Discussion 47
6.1 Experimental results 47
6.1.1 Bacterial segmentation image true positive rate assessment 47
6.1.2 Drug resistance judgment results 61
Chapter 7 Conclusion and Future Work 74
7.1 Conclusion 74
7.2 Future Work 75
References 76

[1].Y.C. Liu, F.Y. Hsu, H.C. Chen, Y.Y. Wang, and Y.N. Sun*, “A Coarse to Fine Auto-Focusing Algorithm for Microscope Image, International Conference on System Science and Engineering, Macao, 2011. (EI)
[2].W.Y. Hsu, W.F.P. Poon, and Y.N. Sun*, “Automatic Seamless Mosaicing of Microscopic Images: Enhancing Appearance with Color Degradation Compensation and Wavelet-based Blending, Journal of Microscopy-Oxford, Vol. 231, No. 3, pp. 408-418, Sep. 2008.
[3].Y.N. Sun*, C.H. Lin, C.C. Kuo, C.L. Ho, and M. C.J. Lin, “Live Cell Tracking Based on Cellular State Recognition from Microscopic Images, Journal of Microscopy-Oxford, Vol. 235, No. 1, pp. 94-105, 2009.
[4].Y.N. Sun*, Y.Y. Wang, S.C. Chang, L.W. Wu, and S.T. Tsai, “Color-Based Tumor Tissue Segmentation for the Automated Estimation of Oral Cancer Parameters, Microscopy Research and Techniques, Vol. 73, Issue 1, pp. 5-13, 2010.
[5].Yi-Ying Wang, Hsin-Chen Chen, Chou-Ching K. Lin, and Yung-Nien Sun*, “Segmentation of Nerve Fiber from Microscopic Cross Sections Based on Coarse-to-fine Registration Strategy, Submitted to Neurocomputing
[6].Y.Y. Wang, C.C.K. Lin, and Y.N. Sun*, “Registration-Based Segmentation of Nerve Cells in Microscopy Images, the 31th Annual International Conf. of the IEEE Engineering in Medicine and Biology Society, pp.6726-6729
[7].Y.C. Liu, H.H. Shih, S.H. Hong, M.J. Huang, S.Y. Huang, and Y.N. Sun*, “Microscopic Image Analysis of Pulley Tissue for Trigger Digits, International Symposium on Clinical Engineering and Medical Informatics暨國科會醫工學門成果發表會, Tainan, Taiwan, 2011.
[8].Y.C. Liu, C.K. Chen, H.C. Chen, S.H. Hong, C.C Yang, I.M. Jou, and Y.N. Sun*, “Quantitative Measurement of Nerve Cells and Myelin Sheaths from Microscopic Images via Two-Staged Segmentation, the 4th Asian Conference on Intelligent Information and Database Systems, Kaohsiung, Taiwan, 2012.
[9].Y.C. Liu, H.H. Shih, T.H. Yang, H.B. Yang, D.S. Yang, and Y.N. Sun*, “Quantitative Measurement for Pathological Change of Pulley Tissue from Microscopic Images via Color-Based Segmentation, the 4th Asian Conference on Intelligent Information and Database Systems, Kaohsiung, Taiwan, 2012.
[10]. N. Otsu, “Threshold Selection Method from Gray-level Histograms, IEEE Transactions on Systems, Man, and Cybernetics, SMC-9, no. 1, pp. 62-66, 1979.
[11].M. Kass, A. Witkin, and D. Terzopoulos, “Snakes: Active Contour Models, International Journal of Computer Vision, 1(4):321–331, 1987
[12].Amir Amini, Terry Weymouth, and Ramesh Jain, “Using dynamic programming for solving variational problems in vision, IEEE Transactions on Pattern Analysis and Machine Intelligence, 12(9):855–867, 1990.
[13].D. Comaniciu, P. Meer, “Mean shift: A Robust Approach Toward Feature Space Analysis, IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 24, no. 5, pp. 603-619, 2002
[14].D. Comaniciu, P. Meer, “Cell Image Segmentation for Diagnostic Pathology, Advanced Algorithmic Approaches to Medical Image Segmentation: State-of-the-art Applications in Cardiology, Neurology, Mammography and Pathology, pp. 541-558, 2001
[15].G. Srinivasa, M. C. Fickus, Y. Guo, A. D. Linstedt, and J. Kovacevic, “Active Mask Segmentation of Fluorescence Microscope Images, IEEE Transactions on Image Processing, vol. 18, no. 8, pp. 1817-1829, 2009.
[16].Xin Li and Michael T. Orchard, “New Edge-Directed Interpolation, IEEE Trans. Image Processing, 2001.
[17].R.C. Gonzalez and R.E. Woods, Digital Image Processing, Prentice Hall, 3rd Edition, 2007.