基于EGEE的医学图像处理若干关键技术研究
|
摘要
医学图像处理与分析技术借助计算机图像处理与分析、计算机图形学、虚拟现实和计算机网络等技术使得医学影像诊疗水平大大提高,日益受到人们的重视,成为国内外研究和应用的热点,并形成了一门新兴的、发展迅速的交叉学科。
随着医学成像技术的发展,医学图像的精度越来越高,数据量也越来越大,因此医学图像处理就需要一种存储容量大运算能力强的处理环境。而EGEE(The Enabling Grids for E-sciencE)恰好能够满足这种需求。因此,研究基于EGEE的医学图像处理技术具有重要意义。
本论文主要研究了以下三方面内容：
设计并开发了基于EGEE的医学图像处理平台,命名为HBB,它基于ITK、VTK和QT而开发,集成了图像滤波、分割和三维可视化等算法,实现了一个功能全面、接口丰富、可扩展性强的医学图像处理平台,可做为一个算法研发的基础开发平台。
针对经典区域增长算法中生长规则确定的困难和单纯使用支持向量机分割速度慢的问题,提出了一种支持向量机与区域增长相结合的图像并行分割方法。首先,从已知分割结果的图像中选取一定数量的目标区域与非目标区域样本点做为支持向量机分类器的训练样本并训练支持向量机,然后利用训练好的支持向量机自动寻找种子点并进行区域增长,在区域增长过程中使用支持向量机分类器作为增长规则,最后,针对边缘和噪声像素点进行必要的后处理。测试实验当使用16结点时分割512×512×64的图像所用时间为132秒,获得了较好的分割效果和较快的分割速度且能实现自动分割,表明所提出的方法是有效可行的。
在医学图像三维重建方面,提出了一种基于EGEE的并行三维重建策略。首先,把任务划分为若干份并分发给子进程；其次,每个子进程采用MC算法提取等值面,将三角面连接成三角带并将结果返回给主进程；最后,主进程将各子进程返回的结果进行合并并显示。实验测试当使用4结点时重建速度提高了约66%,表明该并行三维重建方法能够有效的提高三维重建速度且重建效果良好。
The medical image processing and analysis has improved the level of diagnosis and treat greatly and has been paid more attention to by people with the help of technologies of computer image processing and analysis, computer graphics, virtual reality, computer network and so on. It has become a research and application hotspot at home and abroad, and former a new crossover subject with rapid development.
With the development of the medical imaging technology, the accuracy of medical images is higher and higher, data volume is larger and larger at the same time. Therefore, medical image processing would require a handling environment which has a large storage capacity and a strong computing capable. The EGEE(The Enabling Grids for E-sciencE) just to meet this demand. So the study of medical image processing technology which based on EGEE has great significance.
This thesis includes contents of three main aspects:
Medical image processing platform was designed and developed based on ITK, VTK and QT. It integrated image filtering, segmentation and three-dimensional visualization algorithm and implements a full-featured, interface rich, scalable platform for medical image processing, Can be used as a basis for algorithm development platform for R & D.
In order to solve the difficult of determine the growth rules in conventional regional growth algorithm and the slowly of support vector machine segmentation algorithm, an image segmentation method combined support vector machine and regional growth was proposed. Firstly, select a certain numbers of sample point from target area and non-target area and train the support vector machine classification, then use the trained classification search seed point and regional growing, the support vector machine classification is used as growth rules, the last, some necessary retrogressing were used for the edge and noise. The experimental results show that this algorithm is feasible and it performs better than conventional region growth segmentation algorithm and faster then conventional support vector machine segmentation algorithm.
In the medical image 3D reconstruction, proposed a 3D reconstruction parallel strategy based on EGEE. First, the task is divided into a number of copies and distributed to the sub-processes; secondly, each sub-process using MC algorithm extraction isosurface and linked triangulated surfaces into Triangle and back the results to the main process; finally, the main process will be merge all results of sub-processes returned and displayed. Experimental tests show that the 3D reconstruction parallel method can effectively improve the speed of three-dimensional reconstruction and rebuilding good effect.
随着医学成像技术的发展,医学图像的精度越来越高,数据量也越来越大,因此医学图像处理就需要一种存储容量大运算能力强的处理环境。而EGEE(The Enabling Grids for E-sciencE)恰好能够满足这种需求。因此,研究基于EGEE的医学图像处理技术具有重要意义。
本论文主要研究了以下三方面内容：
设计并开发了基于EGEE的医学图像处理平台,命名为HBB,它基于ITK、VTK和QT而开发,集成了图像滤波、分割和三维可视化等算法,实现了一个功能全面、接口丰富、可扩展性强的医学图像处理平台,可做为一个算法研发的基础开发平台。
针对经典区域增长算法中生长规则确定的困难和单纯使用支持向量机分割速度慢的问题,提出了一种支持向量机与区域增长相结合的图像并行分割方法。首先,从已知分割结果的图像中选取一定数量的目标区域与非目标区域样本点做为支持向量机分类器的训练样本并训练支持向量机,然后利用训练好的支持向量机自动寻找种子点并进行区域增长,在区域增长过程中使用支持向量机分类器作为增长规则,最后,针对边缘和噪声像素点进行必要的后处理。测试实验当使用16结点时分割512×512×64的图像所用时间为132秒,获得了较好的分割效果和较快的分割速度且能实现自动分割,表明所提出的方法是有效可行的。
在医学图像三维重建方面,提出了一种基于EGEE的并行三维重建策略。首先,把任务划分为若干份并分发给子进程；其次,每个子进程采用MC算法提取等值面,将三角面连接成三角带并将结果返回给主进程；最后,主进程将各子进程返回的结果进行合并并显示。实验测试当使用4结点时重建速度提高了约66%,表明该并行三维重建方法能够有效的提高三维重建速度且重建效果良好。
The medical image processing and analysis has improved the level of diagnosis and treat greatly and has been paid more attention to by people with the help of technologies of computer image processing and analysis, computer graphics, virtual reality, computer network and so on. It has become a research and application hotspot at home and abroad, and former a new crossover subject with rapid development.
With the development of the medical imaging technology, the accuracy of medical images is higher and higher, data volume is larger and larger at the same time. Therefore, medical image processing would require a handling environment which has a large storage capacity and a strong computing capable. The EGEE(The Enabling Grids for E-sciencE) just to meet this demand. So the study of medical image processing technology which based on EGEE has great significance.
This thesis includes contents of three main aspects:
Medical image processing platform was designed and developed based on ITK, VTK and QT. It integrated image filtering, segmentation and three-dimensional visualization algorithm and implements a full-featured, interface rich, scalable platform for medical image processing, Can be used as a basis for algorithm development platform for R & D.
In order to solve the difficult of determine the growth rules in conventional regional growth algorithm and the slowly of support vector machine segmentation algorithm, an image segmentation method combined support vector machine and regional growth was proposed. Firstly, select a certain numbers of sample point from target area and non-target area and train the support vector machine classification, then use the trained classification search seed point and regional growing, the support vector machine classification is used as growth rules, the last, some necessary retrogressing were used for the edge and noise. The experimental results show that this algorithm is feasible and it performs better than conventional region growth segmentation algorithm and faster then conventional support vector machine segmentation algorithm.
In the medical image 3D reconstruction, proposed a 3D reconstruction parallel strategy based on EGEE. First, the task is divided into a number of copies and distributed to the sub-processes; secondly, each sub-process using MC algorithm extraction isosurface and linked triangulated surfaces into Triangle and back the results to the main process; finally, the main process will be merge all results of sub-processes returned and displayed. Experimental tests show that the 3D reconstruction parallel method can effectively improve the speed of three-dimensional reconstruction and rebuilding good effect.
引文
[1]GAGLIARDIF, BEGIN M E. EGEE-Providing a Production Quality Grid for E-science[J]. Local to Global Data Interoperability-Challenges and Technologies,2005, (6):88-92.
[2]GAGLIARDI F, GREY F. Old world, new grid[C]. Spectrum, IEEE,2006, 43(7):28-33.
[3]GERMAIN C, BRETON V. Grid-Enabling Medical Image Analysis[J]. Journal of Clinical Monitoring and Computing,2005, (19):339-349.
[4]田捷,包尚联,周明全.医学影像处理与分析[M].北京：电子工业出版社,2003.
[5]THEODOROS E A, George V T, DIMITRA I K. A Grid Enabled Problem Solving Environment for Monte Carlo Matlab Simulations[C]. Distributed Simulation and Real-Time Applications,2007. DS-RT 2007.11th IEEE International Symposium,22-26 Oct.2007:159-166.
[6]APARICIO G, BLANQUER I, HERNANDEZ V. A Parallel Implementation of the K Nearest Neighbors Classifier in Three Levels:Threads, MPI Processes and the Grid[C]. VECPAR 2006, LNCS 4395,2007:225-235.
[7]DIMITROVSKI I, KAKASEVSKI G. Grid Enabled Computer Vision System for Measuring Traffic Parameters[J]. Computing Sciences and Software Engineering,2007,5:561-565.
[8]GERMAIN C, BRETON V, CLARYSSE P. Grid-Enabling Medical Image Analysis[J]. Journal of Clinical Monitoring and Computing,2005,19: 339-349.
[9]GLATARD T, PENNEC X, MONTAGNAT J. Performance Evaluation of Grid Enabled Registration Algorithms Using Bronze-Standards[C]. MICCAI 2006, LNCS 4191,2006:152-160.
[10]CANNAVO F, NUNNARI G, GIORDANO G Variational Method for Image Denoising by Distributed Genetic Algorithms on GRID Environment. Enabling Technologies:Infrastructure for Collaborative Enterprises[C], WETICE'06. 15th IEEE International Workshops on, June 2006:227-232.
[11]金海.搭建资源共享平台：中国教育科研网格ChinaGrid[J].中国教育网络,2006,(9)：25-26.
[12]张翔.支持向量机及其在医学图像分割中的应用[D].武汉：华中科技大学(博士),2004：3-5.
[13]何晓乾.医学图像三维重建技术研究[D].成都：电子科技大学(硕士),2006：2-3.
[14]HERMAN G T, LIU H K. Three dimensional display of human organs from computed tomograms[J]. Computer Graphics and Image Processing,1979,9: 1-21.
[15]GELDER A V, WILHELMS J. Topological considerations in isosurface generation[J]. ACN1 Transactions on Graphics(TOG),1994,13(4):337-375.
[16]LORENSEN W E, CLINE H E. Marching Cubes:a High Resolution 3D Surface Construction Algorithm. Computer Graphics[J].1987,21(4):163-169.
[17]CLINE H E, LORENSEN W E, LUDKE S. Two Algorithms For Three Dimensional Reconstruction of Tomograms[J]. Medical Physics,1988,15(3): 320-327.
[18]DURST M J. Letters:Additional Reference to "Marching cubes"[J]. ACM Computer Graphics,1988,22(4):72-73.
[19]KLEMP J B, MCLRVIN M J, BOYD W S. Poly Paint-a Three Dimensional Rendering Package[C]. In American Meteorology Society Proceedings,6th International Conference on Interactive Animation and Processing Systems, Anaheim, C.A., USA,1990:286-293.
[20]DIO A, KOIDE A. An Efficient Method of Triangulating Equal-valued Surfaces by Using Tetrahedral Cells[J]. IEICE Transactions, E74(1):214-224.
[21]LACROUTE P, LEVOY M. Fast Volume Rendering Using a Shear-Warp Factorization of the Viewing Transformation[J]. Computer Graphics,1994, 24(9):451-457.
[22]WESTOVER L, Footprint Evaluation for Volume Rendering [J]. Computer Graphics,1990:24(4):367-376.
[23]CABRAL B, CAM N. Accelerated Volume Rendering and Homographic Reconstruction using Texture Mapping Hardware[J]. Symposium on Volume Visualization,1994,23(6):91-98.
[24]WERNER M, FRINGS J, WAUQUIEZ F. Topological Design, Routing and Capacity Dimensioning for ISL Networks in Broadband LEO Satellite Systems[J]. Int. J. Sat. Commun.,2001,19(6):499-527.
[25]DONNER A, BERIOLI M. MPLS-based Satellite Constellation Networks[J]. IEEE Journal on Selected Areas in Communications,2004,22(3):598-604.
[26]廖秀秀.多模医学图像处理平台及基于隐含形状表示和边缘信息融合的图像配准算法研究[D].浙江：浙江大学(硕士),2007：10-11.
[27]WILLIAM J S, KENNETH M M, WILLIAM E L. The Design and Implementation of An Object-Oriented Toolkit For 3D Graphics And Visualization[C], Proc. Of IEEE Visualization'96.
[28]BLANCHETTE J, SUMMERFIELD M. C++GUI Programming with Qt 4[M]. Publisher:Prentice Hall Pub Date:June 21,2006:224.
[29]JAYARAM K U. SUPUN S. Fuzzy Connectedness and Object Definition: Theory, Algorithms, and Applications in Image Segmentation[J]. Graphical Models and Image Processing,1996,58(3):246-241.
[30]CORTES C, VAPNIK V. Support Vector Networks[J]. Machine Learning,1995, 20:273-295.
[31]CUI G Q, GAO W. SVMs for Few Examples-based Face Recognition[C]. IEEE ICASSP 2003,2:381-384
[32]马勇,丁晓青.基于层次型支持向量机的人脸检测[J].清华大学学报：自然科学版,2003,43(1)：67-70.
[33]忻栋,杨莹春,吴朝晖.基于SVM-HMM混合模型的说话人确认[J].计算机辅助设计与图形学学报,2003,15(4)：416-420.
[34]高学,金连文,尹俊勋等.一种基于支持向量机的手写汉字识别方法[J].电子学报,2002,30(5)：651-654.
[35]KIMURA Y, WAKAHARA T. Toward Robust Handwritten Kanji Character Recognition [J]. Pattern Recognition Letters,1999,20:979-990.
[36]潘晨,闫相国,郑崇勋等.利用单类支持向量机分割血细胞图像[J].西安交通大学学报,2005,39(2)：150-153.
[37]陈光英,张千里,李星.特征选择和SVM训练模型的联合优化[J].清华大学学报：自然科学版,2004,44(1)：9-12.
[38]祁亨年.支持向量机及其应用研究综述[J].计算机工程,2004,30(10)：6-9.
[39]陈世哲.微电子产品视觉检测中关键技术研究[D].哈尔滨：哈尔滨工业大学(博士),2006：88-92.
[40]HARALICK R M. Statistical and Structural Approaches to Texture[C]. IEEE Proc.1979,67:789-804.
[41]HARALICK R M, SHANMUGAN K, DINSTEIN I. Texture Features for Image Classification[C]. IEEE Trans SMC.1973,8:610-621.
[42]薛志东,隋卫平,李利军.一种SVM与区域生长相结合的图像分割方法[J].计算机应用,2007,27(2)：463-465.
[43]周宴.医学CT图像的三维可视化技术研究[D].重庆：重庆大学(硕士),2005：35-38.
[44]LORENSEN W, CLINE H. Marching Cubes:a high resolution 3D surface construction algorithm[J]. ACM Computer Graphics,1987,21(4):163-170.
[45]梁峰,鲁强,曾绍群.一种基于MPI的并行体绘制算法[J].计算机工程,2005,31(13)：171-173.
[2]GAGLIARDI F, GREY F. Old world, new grid[C]. Spectrum, IEEE,2006, 43(7):28-33.
[3]GERMAIN C, BRETON V. Grid-Enabling Medical Image Analysis[J]. Journal of Clinical Monitoring and Computing,2005, (19):339-349.
[4]田捷,包尚联,周明全.医学影像处理与分析[M].北京：电子工业出版社,2003.
[5]THEODOROS E A, George V T, DIMITRA I K. A Grid Enabled Problem Solving Environment for Monte Carlo Matlab Simulations[C]. Distributed Simulation and Real-Time Applications,2007. DS-RT 2007.11th IEEE International Symposium,22-26 Oct.2007:159-166.
[6]APARICIO G, BLANQUER I, HERNANDEZ V. A Parallel Implementation of the K Nearest Neighbors Classifier in Three Levels:Threads, MPI Processes and the Grid[C]. VECPAR 2006, LNCS 4395,2007:225-235.
[7]DIMITROVSKI I, KAKASEVSKI G. Grid Enabled Computer Vision System for Measuring Traffic Parameters[J]. Computing Sciences and Software Engineering,2007,5:561-565.
[8]GERMAIN C, BRETON V, CLARYSSE P. Grid-Enabling Medical Image Analysis[J]. Journal of Clinical Monitoring and Computing,2005,19: 339-349.
[9]GLATARD T, PENNEC X, MONTAGNAT J. Performance Evaluation of Grid Enabled Registration Algorithms Using Bronze-Standards[C]. MICCAI 2006, LNCS 4191,2006:152-160.
[10]CANNAVO F, NUNNARI G, GIORDANO G Variational Method for Image Denoising by Distributed Genetic Algorithms on GRID Environment. Enabling Technologies:Infrastructure for Collaborative Enterprises[C], WETICE'06. 15th IEEE International Workshops on, June 2006:227-232.
[11]金海.搭建资源共享平台：中国教育科研网格ChinaGrid[J].中国教育网络,2006,(9)：25-26.
[12]张翔.支持向量机及其在医学图像分割中的应用[D].武汉：华中科技大学(博士),2004：3-5.
[13]何晓乾.医学图像三维重建技术研究[D].成都：电子科技大学(硕士),2006：2-3.
[14]HERMAN G T, LIU H K. Three dimensional display of human organs from computed tomograms[J]. Computer Graphics and Image Processing,1979,9: 1-21.
[15]GELDER A V, WILHELMS J. Topological considerations in isosurface generation[J]. ACN1 Transactions on Graphics(TOG),1994,13(4):337-375.
[16]LORENSEN W E, CLINE H E. Marching Cubes:a High Resolution 3D Surface Construction Algorithm. Computer Graphics[J].1987,21(4):163-169.
[17]CLINE H E, LORENSEN W E, LUDKE S. Two Algorithms For Three Dimensional Reconstruction of Tomograms[J]. Medical Physics,1988,15(3): 320-327.
[18]DURST M J. Letters:Additional Reference to "Marching cubes"[J]. ACM Computer Graphics,1988,22(4):72-73.
[19]KLEMP J B, MCLRVIN M J, BOYD W S. Poly Paint-a Three Dimensional Rendering Package[C]. In American Meteorology Society Proceedings,6th International Conference on Interactive Animation and Processing Systems, Anaheim, C.A., USA,1990:286-293.
[20]DIO A, KOIDE A. An Efficient Method of Triangulating Equal-valued Surfaces by Using Tetrahedral Cells[J]. IEICE Transactions, E74(1):214-224.
[21]LACROUTE P, LEVOY M. Fast Volume Rendering Using a Shear-Warp Factorization of the Viewing Transformation[J]. Computer Graphics,1994, 24(9):451-457.
[22]WESTOVER L, Footprint Evaluation for Volume Rendering [J]. Computer Graphics,1990:24(4):367-376.
[23]CABRAL B, CAM N. Accelerated Volume Rendering and Homographic Reconstruction using Texture Mapping Hardware[J]. Symposium on Volume Visualization,1994,23(6):91-98.
[24]WERNER M, FRINGS J, WAUQUIEZ F. Topological Design, Routing and Capacity Dimensioning for ISL Networks in Broadband LEO Satellite Systems[J]. Int. J. Sat. Commun.,2001,19(6):499-527.
[25]DONNER A, BERIOLI M. MPLS-based Satellite Constellation Networks[J]. IEEE Journal on Selected Areas in Communications,2004,22(3):598-604.
[26]廖秀秀.多模医学图像处理平台及基于隐含形状表示和边缘信息融合的图像配准算法研究[D].浙江：浙江大学(硕士),2007：10-11.
[27]WILLIAM J S, KENNETH M M, WILLIAM E L. The Design and Implementation of An Object-Oriented Toolkit For 3D Graphics And Visualization[C], Proc. Of IEEE Visualization'96.
[28]BLANCHETTE J, SUMMERFIELD M. C++GUI Programming with Qt 4[M]. Publisher:Prentice Hall Pub Date:June 21,2006:224.
[29]JAYARAM K U. SUPUN S. Fuzzy Connectedness and Object Definition: Theory, Algorithms, and Applications in Image Segmentation[J]. Graphical Models and Image Processing,1996,58(3):246-241.
[30]CORTES C, VAPNIK V. Support Vector Networks[J]. Machine Learning,1995, 20:273-295.
[31]CUI G Q, GAO W. SVMs for Few Examples-based Face Recognition[C]. IEEE ICASSP 2003,2:381-384
[32]马勇,丁晓青.基于层次型支持向量机的人脸检测[J].清华大学学报：自然科学版,2003,43(1)：67-70.
[33]忻栋,杨莹春,吴朝晖.基于SVM-HMM混合模型的说话人确认[J].计算机辅助设计与图形学学报,2003,15(4)：416-420.
[34]高学,金连文,尹俊勋等.一种基于支持向量机的手写汉字识别方法[J].电子学报,2002,30(5)：651-654.
[35]KIMURA Y, WAKAHARA T. Toward Robust Handwritten Kanji Character Recognition [J]. Pattern Recognition Letters,1999,20:979-990.
[36]潘晨,闫相国,郑崇勋等.利用单类支持向量机分割血细胞图像[J].西安交通大学学报,2005,39(2)：150-153.
[37]陈光英,张千里,李星.特征选择和SVM训练模型的联合优化[J].清华大学学报：自然科学版,2004,44(1)：9-12.
[38]祁亨年.支持向量机及其应用研究综述[J].计算机工程,2004,30(10)：6-9.
[39]陈世哲.微电子产品视觉检测中关键技术研究[D].哈尔滨：哈尔滨工业大学(博士),2006：88-92.
[40]HARALICK R M. Statistical and Structural Approaches to Texture[C]. IEEE Proc.1979,67:789-804.
[41]HARALICK R M, SHANMUGAN K, DINSTEIN I. Texture Features for Image Classification[C]. IEEE Trans SMC.1973,8:610-621.
[42]薛志东,隋卫平,李利军.一种SVM与区域生长相结合的图像分割方法[J].计算机应用,2007,27(2)：463-465.
[43]周宴.医学CT图像的三维可视化技术研究[D].重庆：重庆大学(硕士),2005：35-38.
[44]LORENSEN W, CLINE H. Marching Cubes:a high resolution 3D surface construction algorithm[J]. ACM Computer Graphics,1987,21(4):163-170.
[45]梁峰,鲁强,曾绍群.一种基于MPI的并行体绘制算法[J].计算机工程,2005,31(13)：171-173.