肾小球三维重建系统的研究
|
摘要
计算机技术和数字图像处理技术的不断发展，使得医学图像诊断在现代医疗中的地位越来越重要。现代医学已经越来越离不开医学图像所提供的信息，医学图像在临床诊断、教学科研等方面有着极其重要的作用。目前，已经成为提高疾病的早期诊断，以及更准确的分析病情的不可缺少的寻征手段。
本文针对肾脏组织切片图像中二维肾小球的提取及三维模型的建立显示，设计了一个可以自动进行二维提取和三维重建的辅助系统。
文中首先对原始医学图像进行读取，通过预处理解决图像噪声以及光照不均等问题，获得较清晰的二维切片图像。然后，通过一种基于轮廓信息的区域增长方法，对所建立的集合数据进行肾小球区域的提取，最后对配准后的所有二维肾小球图像建立体数据集，进行可视化成像处理。
其中，在对二维序列图像进行肾小球区域提取时，本文采用了基于活动轮廓的方法，获得序列切片图像中的肾小球的大致边界，再对该边界的一个小区域内作区域生长，最终得到真实的包围肾小球区域的闭合边界。
然后，以提取出的肾小球的边界为基本，往相邻的组织切片图像上投影，再进行GVF轮廓提取和区域生长分割，最终提取出序列图像中所有的二维肾小球区域。在此过程中，都是以上次提取出的肾小球边界为下次提取的初始轮廓，以减少累计误差。
在三维可视化处理当中，本文根据三维数据体本身的属性和光线跟踪的原理，进行了基于体素的直接体绘制，并利用二维提取出的肾小球图像，进行了面绘制，实现了肾小球的外貌特征。
实验表明，本文所提出的方法对于医学肾脏切片图像，二维提取和三维重建有很好的效果，可以作为医学辅助诊断软件的一个前期实验软件，基本满足了要求。
With the continuous development of computer technology and digital image processing technology, medical image diagnosis plays a more and more role in modern medical treatment. Modern medicine has become increasingly inseparable from the information provided by medical image, which makes a difference in clinic diagnosis as well as in teaching and scientific research. Currently, it becomes an indispensable method to improve the early diagnosis and diagnose diseases accurately.
In this essay, in view of two-dimensional glomerulus extraction and three-dimensional model establishment display in the sliced image of kidney tissue, an assistant system is designed which is able to carry on automatic two-dimensional extraction and three-dimensional re-establishment. At first, a clear two-dimensional sliced image is got by reading the original medical image and wiping off noise, balancing the brightness with preterit. Secondly, through a way of area increase based on boundary information, two-dimensional extraction is made on the established set data. Finally, all the matched glomerular will be collected to establish a volume data set, carrying on visualized imaging.
Therein, while carrying on extraction on the two-dimensional images, in this thesis, the active outline model is used to extract main glomerulus borderline in a sliced image. Then, carry on regional growth base the borderline in a small area. At last, the true close borderline of the glomerulus is attained.
And then, based on glomerular boundary extracted, project on adjacent image, do again the GVF borderline extraction and the regional growth, finally all two-dimensional glomerular areas are extracted. In this process the extracted glomerular boundary will be the initialization borderline for the next extraction so as to reduce the accumulating error. In three-dimensional visualized process, in light of attribute of three-dimensional data and ray tracing principle, direct volume rendering is carried out. And with extracted glomerulus, surface rendering is carried on. As a result of this, exterior feature of glomerulus of kidney is realized.
It can be seen from the experiment that the way put forward in this essay is of great significance for sliced image of medical kidney, two-dimensional extraction and three-dimensional re-establishment. Accordingly, it can be used as pre-experimental software of medical assistant diagnosis software, meeting the basic requirement.
本文针对肾脏组织切片图像中二维肾小球的提取及三维模型的建立显示，设计了一个可以自动进行二维提取和三维重建的辅助系统。
文中首先对原始医学图像进行读取，通过预处理解决图像噪声以及光照不均等问题，获得较清晰的二维切片图像。然后，通过一种基于轮廓信息的区域增长方法，对所建立的集合数据进行肾小球区域的提取，最后对配准后的所有二维肾小球图像建立体数据集，进行可视化成像处理。
其中，在对二维序列图像进行肾小球区域提取时，本文采用了基于活动轮廓的方法，获得序列切片图像中的肾小球的大致边界，再对该边界的一个小区域内作区域生长，最终得到真实的包围肾小球区域的闭合边界。
然后，以提取出的肾小球的边界为基本，往相邻的组织切片图像上投影，再进行GVF轮廓提取和区域生长分割，最终提取出序列图像中所有的二维肾小球区域。在此过程中，都是以上次提取出的肾小球边界为下次提取的初始轮廓，以减少累计误差。
在三维可视化处理当中，本文根据三维数据体本身的属性和光线跟踪的原理，进行了基于体素的直接体绘制，并利用二维提取出的肾小球图像，进行了面绘制，实现了肾小球的外貌特征。
实验表明，本文所提出的方法对于医学肾脏切片图像，二维提取和三维重建有很好的效果，可以作为医学辅助诊断软件的一个前期实验软件，基本满足了要求。
With the continuous development of computer technology and digital image processing technology, medical image diagnosis plays a more and more role in modern medical treatment. Modern medicine has become increasingly inseparable from the information provided by medical image, which makes a difference in clinic diagnosis as well as in teaching and scientific research. Currently, it becomes an indispensable method to improve the early diagnosis and diagnose diseases accurately.
In this essay, in view of two-dimensional glomerulus extraction and three-dimensional model establishment display in the sliced image of kidney tissue, an assistant system is designed which is able to carry on automatic two-dimensional extraction and three-dimensional re-establishment. At first, a clear two-dimensional sliced image is got by reading the original medical image and wiping off noise, balancing the brightness with preterit. Secondly, through a way of area increase based on boundary information, two-dimensional extraction is made on the established set data. Finally, all the matched glomerular will be collected to establish a volume data set, carrying on visualized imaging.
Therein, while carrying on extraction on the two-dimensional images, in this thesis, the active outline model is used to extract main glomerulus borderline in a sliced image. Then, carry on regional growth base the borderline in a small area. At last, the true close borderline of the glomerulus is attained.
And then, based on glomerular boundary extracted, project on adjacent image, do again the GVF borderline extraction and the regional growth, finally all two-dimensional glomerular areas are extracted. In this process the extracted glomerular boundary will be the initialization borderline for the next extraction so as to reduce the accumulating error. In three-dimensional visualized process, in light of attribute of three-dimensional data and ray tracing principle, direct volume rendering is carried out. And with extracted glomerulus, surface rendering is carried on. As a result of this, exterior feature of glomerulus of kidney is realized.
It can be seen from the experiment that the way put forward in this essay is of great significance for sliced image of medical kidney, two-dimensional extraction and three-dimensional re-establishment. Accordingly, it can be used as pre-experimental software of medical assistant diagnosis software, meeting the basic requirement.
引文
[1] 谢院生，井口清太郎，西慎一，下条文武．99日本肾脏周概况．肾脏病与透析肾移植杂志，9(1)，2000：95-97．
[2] 邹万忠，谌贻璞，章友康等．肾活检病理诊断图鉴．北京：人民卫生出版社，2000．
[3] 刘志红，黎磊石．迈向21世纪的中国肾脏病临床。肾脏病与透析肾移植杂志，10(1)，2001：1-2．
[4] 王莉，叶一舟，廖洪军等．从石蜡包埋的微量肾活检组织中提取及分析RNA．中国组织化学与细胞化学杂志，5(3)，1996：346-350．
[5] 胡晓莉．做好肾活检组织制片技术的几点体会．内蒙古医学院学报，20(4)，1998：283-284．
[6] 许朝晖毕业设计论文．医学图像识别技术的研究．2003．
[7] 张军硕士论文．肾小球提取及球内细胞核的统计分析．2004．
[8] 张小牤博士论文．医学图像的领域分割法．1996．
[9] 朱虹博士论文．肾脏组织图像的解析方法的研究．1999．
[10] 郑南宁．计算机视觉与模式识别．北京：国防工业出版社，1998．3．
[11] 谷口庆治编，朱虹等译．数字图像处理—应用篇．北京：科学出版社，2002．7．
[12] 袁华，滕奇志，袁支润，陶德元．边缘检测及其在医学图像中的应用．生物医学工程学杂志，18(1)，2001：149-153．
[13] J. FMangin, V. Frouin, I. Bloch, J, Regis, J. Lopez-Krahe。From 3Dmangetic resonance images to structural representations of the cortex topography using to[ology preserving deformations Math. imag Vis 1995, 5: 297～318
[14] Kikinis Statistical intensity correction and segmentation of magnetic resonance image data 1994. P13-14
[15] Dong Joong Kang. A Fast and Stable Snake Algorithm for Medical Images. Pattern Recognition Letters, 20, 1999: 507-512.
[16] 赵巨波，孙华燕，杜巍．一种图像边缘特征提取算法．光学精密工程，8(4)，2000：325—327．
[17] Fan JP, Yau DKY, Elmagarmid AK et al. Automatic Image Segmentation by Integrating Color-edge Extraction and Seeded Region Growing. IEEE Transactions on Image Processing, 10(10), 2001: 1454—1461.
[18] Tsai CM, Lee HJ, Binarization of Color Document Images via Luminance and Saturation Color Features. IEEE Transactions on Image Processing, 11(4), 2002:434—439.
[19] Zhu H, Taniguchi K, Watanabe SZ, et al. Extraction of Renal Glomerular Regions Using a Non-linear Difference Filter and Region-estimation Model. MEDICAL IMAGING TECHNOLOGY. 17(6), 1999: 669—773.
[20] 张军，朱虹，许朝晖等．神经元网络容错学习在肾小球区域边界增强中的应朋．生物医学工程学报，2006 Vol．23 NO．3 P653-656．
[21] 基于多分辨率小波和高斯混合模型的纹理图像分割Textured Image Segmentation Based on Multiresolution Wavelet and Gaussian Mixture Models[北京大学学报(自然科学版)Acta Scientiarum Naturalium Universitatis Pekinensis]余鹏，封举富，YU Peng，FENG Jufu
[22] 王月兰，曾迎生．信息融合技术在彩色图像分割方法中的应用，计算机学报．23(7)，2002：763-767．
[23] Zhu B, Taniguchi K. A method of binalization for renal images using the self organized neural network. Trans IEE Japan. I18-C(9) 1998: 1397-1398.
[24] 姚力，刘佳敏，谢咏圭等．基于细胞神经网络的图像分割及其在医学图像中的应用[J]，中国科学．31(2)，2001：1
[25] Chow CK, Kaneko T. Boundary detection of radiographic images by a thresholdmethod[M]. New York, Academic 1972
[26] Clark J. Authenticating Edges Produced by Zero-Crossing Algorithm. IEEE Trans on Pattern Analysis and Machine Intelligence, 1989, 11(1): 4357
[27] Lu Y, Jain R C. Behavior of Edges in Scale Space. IEEE Trans on Pattern Analysis and Machine Intelligence, 1988, 11(4): 337356
[28] Lee J S. Morphological edge detection. IEEE Journal of Robotics and Automation, 1987, 3(2): 5978
[29] Chen Wunfan, Lu Xia. A new algorithm of edge detection for color image: generalized fussyoperator. Science in China(A), 1995, 38(10): 265272
[30] Chandrajit L. Bajaj, Valerio Pascucci, and Daniel R, Schilore, the contour spectrum In Proceeding Visualization' 97 pages 167-173 1997
[31] Falcao A X Udupa J K, Smarasekera S Sharma S. User-Stewed Image SegmentationParadigms: Live Wire and Live Lane. Graphic Model and Image Processing, 1998, 60(4); 233260
[32] Mclnerney T, Terzopoulos D. Defonnable Models in Medical Image Analysis: ASurvey. Medical ImageAnalysis, 1996 4(2): 91108
[33] Terzopoulos D, Fleischer K. Deformable models. The Visual Computer 1988, 4: 306-331
[34] Kass M Witkin A Terzopoulous D. Snakes: Active Contour Models. International Journal of Computer Vsion 1988 1(4): 321331
[35] Yuille A, Hallinan P. Deformable templates. Active Vision, 1992: 2038
[36] Terzopoulos D, Water K. Analysis of facial images using physical and anatomical models. ComputerVision, 1990: 727732
[37] Cohen L D, Cohen L, Finite-element methods for active contour models and balloons for 2-D and 3-Dimages. IEEE Transactions on Pattern Analysis and Machine Intelligence. 1993. 11: 11311147
[38] Amini A A, Weymouth T E, Jain R C. Using dynamic programming for solving variational problems invision. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1990, 9, 12(9): 855867
[39] Williams J D, Shah M. A fast algorithm for active contours and curvature estimation. CVGIP: ImageUnderstanding, 1992, 55(1): 1426
[40] Schnabei JA, Arridge SR. Active contour models for shape description using multiscale differential in variants, BMVC95 Vol 1, 1995: 197206
[41] 唐泽圣，三维数据场可视化{M}．北京：清华大学出版社，1999。
[42] 钟世镇，数字化虚拟人体研究现状和展望，解放军医学杂志，2003。
[43] A. C. Martins, E. B. Pires, L. R. Salvado, etal. Anumerical model of the passive and active behavior of skeletal muscles[J]. Computer Methods in Applied Mechanics and Engineering, 1998, 151, 419-433.
[44] W. lorensen and H. Cline Marching Cubes:a High resolutiong 3D surface construction algorithm. ACM Computer Graphics 1987. 21(4):163-170.
[2] 邹万忠，谌贻璞，章友康等．肾活检病理诊断图鉴．北京：人民卫生出版社，2000．
[3] 刘志红，黎磊石．迈向21世纪的中国肾脏病临床。肾脏病与透析肾移植杂志，10(1)，2001：1-2．
[4] 王莉，叶一舟，廖洪军等．从石蜡包埋的微量肾活检组织中提取及分析RNA．中国组织化学与细胞化学杂志，5(3)，1996：346-350．
[5] 胡晓莉．做好肾活检组织制片技术的几点体会．内蒙古医学院学报，20(4)，1998：283-284．
[6] 许朝晖毕业设计论文．医学图像识别技术的研究．2003．
[7] 张军硕士论文．肾小球提取及球内细胞核的统计分析．2004．
[8] 张小牤博士论文．医学图像的领域分割法．1996．
[9] 朱虹博士论文．肾脏组织图像的解析方法的研究．1999．
[10] 郑南宁．计算机视觉与模式识别．北京：国防工业出版社，1998．3．
[11] 谷口庆治编，朱虹等译．数字图像处理—应用篇．北京：科学出版社，2002．7．
[12] 袁华，滕奇志，袁支润，陶德元．边缘检测及其在医学图像中的应用．生物医学工程学杂志，18(1)，2001：149-153．
[13] J. FMangin, V. Frouin, I. Bloch, J, Regis, J. Lopez-Krahe。From 3Dmangetic resonance images to structural representations of the cortex topography using to[ology preserving deformations Math. imag Vis 1995, 5: 297～318
[14] Kikinis Statistical intensity correction and segmentation of magnetic resonance image data 1994. P13-14
[15] Dong Joong Kang. A Fast and Stable Snake Algorithm for Medical Images. Pattern Recognition Letters, 20, 1999: 507-512.
[16] 赵巨波，孙华燕，杜巍．一种图像边缘特征提取算法．光学精密工程，8(4)，2000：325—327．
[17] Fan JP, Yau DKY, Elmagarmid AK et al. Automatic Image Segmentation by Integrating Color-edge Extraction and Seeded Region Growing. IEEE Transactions on Image Processing, 10(10), 2001: 1454—1461.
[18] Tsai CM, Lee HJ, Binarization of Color Document Images via Luminance and Saturation Color Features. IEEE Transactions on Image Processing, 11(4), 2002:434—439.
[19] Zhu H, Taniguchi K, Watanabe SZ, et al. Extraction of Renal Glomerular Regions Using a Non-linear Difference Filter and Region-estimation Model. MEDICAL IMAGING TECHNOLOGY. 17(6), 1999: 669—773.
[20] 张军，朱虹，许朝晖等．神经元网络容错学习在肾小球区域边界增强中的应朋．生物医学工程学报，2006 Vol．23 NO．3 P653-656．
[21] 基于多分辨率小波和高斯混合模型的纹理图像分割Textured Image Segmentation Based on Multiresolution Wavelet and Gaussian Mixture Models[北京大学学报(自然科学版)Acta Scientiarum Naturalium Universitatis Pekinensis]余鹏，封举富，YU Peng，FENG Jufu
[22] 王月兰，曾迎生．信息融合技术在彩色图像分割方法中的应用，计算机学报．23(7)，2002：763-767．
[23] Zhu B, Taniguchi K. A method of binalization for renal images using the self organized neural network. Trans IEE Japan. I18-C(9) 1998: 1397-1398.
[24] 姚力，刘佳敏，谢咏圭等．基于细胞神经网络的图像分割及其在医学图像中的应用[J]，中国科学．31(2)，2001：1
[25] Chow CK, Kaneko T. Boundary detection of radiographic images by a thresholdmethod[M]. New York, Academic 1972
[26] Clark J. Authenticating Edges Produced by Zero-Crossing Algorithm. IEEE Trans on Pattern Analysis and Machine Intelligence, 1989, 11(1): 4357
[27] Lu Y, Jain R C. Behavior of Edges in Scale Space. IEEE Trans on Pattern Analysis and Machine Intelligence, 1988, 11(4): 337356
[28] Lee J S. Morphological edge detection. IEEE Journal of Robotics and Automation, 1987, 3(2): 5978
[29] Chen Wunfan, Lu Xia. A new algorithm of edge detection for color image: generalized fussyoperator. Science in China(A), 1995, 38(10): 265272
[30] Chandrajit L. Bajaj, Valerio Pascucci, and Daniel R, Schilore, the contour spectrum In Proceeding Visualization' 97 pages 167-173 1997
[31] Falcao A X Udupa J K, Smarasekera S Sharma S. User-Stewed Image SegmentationParadigms: Live Wire and Live Lane. Graphic Model and Image Processing, 1998, 60(4); 233260
[32] Mclnerney T, Terzopoulos D. Defonnable Models in Medical Image Analysis: ASurvey. Medical ImageAnalysis, 1996 4(2): 91108
[33] Terzopoulos D, Fleischer K. Deformable models. The Visual Computer 1988, 4: 306-331
[34] Kass M Witkin A Terzopoulous D. Snakes: Active Contour Models. International Journal of Computer Vsion 1988 1(4): 321331
[35] Yuille A, Hallinan P. Deformable templates. Active Vision, 1992: 2038
[36] Terzopoulos D, Water K. Analysis of facial images using physical and anatomical models. ComputerVision, 1990: 727732
[37] Cohen L D, Cohen L, Finite-element methods for active contour models and balloons for 2-D and 3-Dimages. IEEE Transactions on Pattern Analysis and Machine Intelligence. 1993. 11: 11311147
[38] Amini A A, Weymouth T E, Jain R C. Using dynamic programming for solving variational problems invision. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1990, 9, 12(9): 855867
[39] Williams J D, Shah M. A fast algorithm for active contours and curvature estimation. CVGIP: ImageUnderstanding, 1992, 55(1): 1426
[40] Schnabei JA, Arridge SR. Active contour models for shape description using multiscale differential in variants, BMVC95 Vol 1, 1995: 197206
[41] 唐泽圣，三维数据场可视化{M}．北京：清华大学出版社，1999。
[42] 钟世镇，数字化虚拟人体研究现状和展望，解放军医学杂志，2003。
[43] A. C. Martins, E. B. Pires, L. R. Salvado, etal. Anumerical model of the passive and active behavior of skeletal muscles[J]. Computer Methods in Applied Mechanics and Engineering, 1998, 151, 419-433.
[44] W. lorensen and H. Cline Marching Cubes:a High resolutiong 3D surface construction algorithm. ACM Computer Graphics 1987. 21(4):163-170.