基于支持向量机的肺部结节CT图像分割与识别
|
摘要
基于支持向量机的肺部结节计算机断层扫描(Computed Tomography，CT）图像的分割与识别是肺部计算机辅助诊疗(Computer-Aid Diagnosis，CAD）的核心，能够提供精简准确信息，提高医师诊断效率。准确的图像分割方法、合理的图像特征提取和泛化能力较强的机器学习方法是计算机辅助诊疗的关键。针对图像分割与识别问题的国内外研究重点如下：
1.肺部结节提取方面，研究重点是肺结节的特性分析。根据图像的特点和肺部结节的特点使用针对性较强的方法分别提取孤立型肺部结节和粘连肺壁的肺部结节。
2.肿瘤识别方面，研究重点是如何选择合适的图像特征，通过统计学理论，设计期望风险尽可能小的识别分类系统。
针对以上问题，本文的主要研究工作如下：
1.根据计算机断层扫描图像的成像特点和肺部图像的特点，使用基于改进滚球法的分割方法对粘连肺壁的肺部结节区域进行提取；使用区域增长法对肺区内孤立型结节区域进行提取。提取结果作为后续识别的学习样本库。实验结果表明，改进的滚球法可以有效提取粘连肺壁的结节，基于区域增长的方法可以有效提取独立型结节。
2.提出了一种将基于规则的对感兴趣区域（Region of Interest，ROI）进行初提取的方法与最小二乘支持向量机（Least Squares Support Vector Machine，LSSVM）方法相结合的肺部结节CT图像识别方法。通过对图像中的提取结节的图像特征分析和对肺部结节区域的先验知识的分析，首先使用基于特定规则的样本提取方法，排除了一定数量的非结节感兴趣区域，然后使用基于最小二乘支持向量机的方法对其识别，从而实现了样本由低维空间到高维空间的映射，有效的避免了人工神经网络（Artificial Neural Network，ANN）等方法造成的局部最优解、高维运算复杂、泛化能力差和过学习等问题。实验结果表明，合理的选取图像特征后，使用最小二乘支持向量机的方法可以在少量增加漏检数的情况下，降低误判数。
The segmentation and recognition of lung nodules in computed tomography (CT)images based on the support vector machine is the core of the computer-aideddiagnosis (CAD).The CAD can provide concise and accurate information andimprove the efficiency of medical diagnosis. The accurate image segmentationmethod, the reasonable image feature extraction and the powerful generalizationability of the machine learning are the key of the computer-aided diagnosis. The mainpoint of the segmentation and recognition of images is introduced as follows:
1. In the segmentation and extraction of lung nodules aspects, the characteristicsanalysis of lung nodules section is the key in research. According to the features of theimage and the lung nodules, the highly targeted methods are used to extract thesolitary lung nodules and the adhesion lung wall of lung nodules respectively.
2. In the tumor recognition aspects, the research focuses on how to reasonablychoose the appropriate image features. By the statistical theory, the identification andclassification system which expected risky is as small as possible is designed.
For the above problems, the main research is as follows:
1. According to the characteristics of the computer tomography image and lungnodules images, a method of image segmentation based on the improved rolling ballis used to extract the adhesion lung wall of lung nodules region; the region growingmethod is used to extract the independent nodules region in the lung area. Theextraction results could be a study sample database for the follow-up identification.The experimental results show that the improved rolling ball method can extractnodules which adhesive the wall of lung and the methods based on region growingcan extract the independent nodules.
2. The recognition method for CT images of lung nodules combined the earlyregion of interest (ROI) extraction based on the rule with the least square supportvector machine (LSSVM) is proposed. By the analysis of the characteristics of theextraction nodules and the prior knowledge of the lung nodules, the sample extractionmethod based on the specific rules is used, and the certain number of the non-nodulesregion of interest is excluded. Then the method based on the least square supportvector machine is used for recognition to achieve the mapping which is from lowdimension space to high dimension space. So the problems of the local optimalsolutions, the complex high-dimensional operation, the poor generalization ability andthe over learning caused by the artificial neural network are avoided effectively. Theexperimental results show that the least squares support vector machine could reducethe false positives after the reasonable selection of the image characters when thefalse negatives are increased in a small amount.
1.肺部结节提取方面，研究重点是肺结节的特性分析。根据图像的特点和肺部结节的特点使用针对性较强的方法分别提取孤立型肺部结节和粘连肺壁的肺部结节。
2.肿瘤识别方面，研究重点是如何选择合适的图像特征，通过统计学理论，设计期望风险尽可能小的识别分类系统。
针对以上问题，本文的主要研究工作如下：
1.根据计算机断层扫描图像的成像特点和肺部图像的特点，使用基于改进滚球法的分割方法对粘连肺壁的肺部结节区域进行提取；使用区域增长法对肺区内孤立型结节区域进行提取。提取结果作为后续识别的学习样本库。实验结果表明，改进的滚球法可以有效提取粘连肺壁的结节，基于区域增长的方法可以有效提取独立型结节。
2.提出了一种将基于规则的对感兴趣区域（Region of Interest，ROI）进行初提取的方法与最小二乘支持向量机（Least Squares Support Vector Machine，LSSVM）方法相结合的肺部结节CT图像识别方法。通过对图像中的提取结节的图像特征分析和对肺部结节区域的先验知识的分析，首先使用基于特定规则的样本提取方法，排除了一定数量的非结节感兴趣区域，然后使用基于最小二乘支持向量机的方法对其识别，从而实现了样本由低维空间到高维空间的映射，有效的避免了人工神经网络（Artificial Neural Network，ANN）等方法造成的局部最优解、高维运算复杂、泛化能力差和过学习等问题。实验结果表明，合理的选取图像特征后，使用最小二乘支持向量机的方法可以在少量增加漏检数的情况下，降低误判数。
The segmentation and recognition of lung nodules in computed tomography (CT)images based on the support vector machine is the core of the computer-aideddiagnosis (CAD).The CAD can provide concise and accurate information andimprove the efficiency of medical diagnosis. The accurate image segmentationmethod, the reasonable image feature extraction and the powerful generalizationability of the machine learning are the key of the computer-aided diagnosis. The mainpoint of the segmentation and recognition of images is introduced as follows:
1. In the segmentation and extraction of lung nodules aspects, the characteristicsanalysis of lung nodules section is the key in research. According to the features of theimage and the lung nodules, the highly targeted methods are used to extract thesolitary lung nodules and the adhesion lung wall of lung nodules respectively.
2. In the tumor recognition aspects, the research focuses on how to reasonablychoose the appropriate image features. By the statistical theory, the identification andclassification system which expected risky is as small as possible is designed.
For the above problems, the main research is as follows:
1. According to the characteristics of the computer tomography image and lungnodules images, a method of image segmentation based on the improved rolling ballis used to extract the adhesion lung wall of lung nodules region; the region growingmethod is used to extract the independent nodules region in the lung area. Theextraction results could be a study sample database for the follow-up identification.The experimental results show that the improved rolling ball method can extractnodules which adhesive the wall of lung and the methods based on region growingcan extract the independent nodules.
2. The recognition method for CT images of lung nodules combined the earlyregion of interest (ROI) extraction based on the rule with the least square supportvector machine (LSSVM) is proposed. By the analysis of the characteristics of theextraction nodules and the prior knowledge of the lung nodules, the sample extractionmethod based on the specific rules is used, and the certain number of the non-nodulesregion of interest is excluded. Then the method based on the least square supportvector machine is used for recognition to achieve the mapping which is from lowdimension space to high dimension space. So the problems of the local optimalsolutions, the complex high-dimensional operation, the poor generalization ability andthe over learning caused by the artificial neural network are avoided effectively. Theexperimental results show that the least squares support vector machine could reducethe false positives after the reasonable selection of the image characters when thefalse negatives are increased in a small amount.
引文
[1]杨玉海.孤立性肺结节良恶性综合性影像诊断的ROC分析及其临床应用价值研究[D].山东大学，2011.
[2]梁琰.肺部CT图像分析及特征提取研究[D].重庆大学，2007.
[3]孙旭辉.基于CT图像的肺结节检测与识别算法研究[D].太原科技大学，2011.
[4]杨合超，周雪梅.几种图像分割技术的比较[J].电脑知识与技术，2009(9):2440-2441.
[5]杨怀义.图像分割中算法的应用研究[J].计算机仿真，2012(02):229-232.
[6]胡敏，李梅，汪荣贵.改进的Otsu算法在图像分割中的应用[J].电子测量与仪器学报，2010(05):443-449.
[7]楚存坤，李月卿，王昌元.医学图像的分割技术及其新进展[N].泰山医学院学报，2007,28(4):315-317.
[8] Samuel G. A., Maryellen L. G., Catherine J. M., et al. Computerized Detection ofPulmonary Nodules on CT Scans. Imaging&Therapeutic Technology[J].1999,19:1303-1311.
[9] Gurcan M. N., Sanhiner B., Petrick N. Lung Nodule Detection on ThoracicComputed Tomography Images: Preliminary Evaluation of a Computer-aidedDiagnosis Systems [J]. Med Phys,2002,29:2552-2558.
[10] M. Dajnowiec, J. Alirezaie. Computer Simulation for Segmentation of LungNodules in CT Images[C]. International Conference on Systems: Man andCybernetics. Toronto,2004:4491~4496.
[11] J.M. Kuhnigk, J. Dicken, L. Bornemann, et al. Morphological Segmentation andPartial Volume Analysis for Volumetry of Solid Pulmonary Lesions in ThoracicCT Scans[C]. IEEE Transactions on Medical Imaging,2006,25(4):417~434.
[12] Lee Y., Hara T., Fujita H., et al. Automated Detection of Pulmonary Nodules inHelical CT Images based on an Improved Template-Matching Technique[J].IEEE Trans. Med. Imaging.2001,20:595–604.
[13] Okumura T, Miwa, Jun-ichi T, Kako, et al. Automatic Detection of Lung CancerChest CT Images by Variable N-Quoit Filter[C].Fourteenth InternationalConference on Pattern Recognition,1998,2(8):1671–1673.
[14] Carol L N, Li F. An interactive system for CT lung nodule identification andexamination[C].Elsevier Science International Congress,2001,1230:639-645.
[15] Li F, Carol L N, Jianzhong Q, et al. Automatic detection of lung nodules frommulti-slice low-dose CT images[J].Proc. SPIE2001,4322:1828-1835.
[16] Y. Yim, H. Hong, Y. G. Shin. Hybrid Lung Segmentation in Chest CT Images forComputer-Aided Diagnosis[C]. Proceedings of7th International Workshop onEnterprise networking and Computing in Healthcare Industry. Kuala,2005:378~383.
[17] Fetita C. I., Prêteux F. J., Beigelman-Aubry C., et al.3-D Automated LungNodule Segmentation in HRCT[J]. Lecture Notes in Computer Science,2003,2878/2003:626–634.
[18]吴龙海.高分辨率CT图像的肺部病变计算机辅助诊断研究[D].中国科学技术大学，2008.
[19]孙申申.基于CT影像的肺结节检测与分割方法研究[D].东北大学，2009.
[20]王倩. CT图像中肺部疾病的计算机辅助诊断方法研究[D].华中科技大学，2009
[21] Suzuki K, Feng Li, Sone S, Doi K. Computer-aided diagnostic scheme fordistinction between benign and malignant nodules in thoracic low-dose CT by useof massive training artificial neural network[C]. IEEE Transactions on MedicalImage,2005,24:1138–1150.
[22] Lin DT, Yan CR, Chen WT. Autonomous detection of pulmonary nodules on CTimages with aneural network-based fuzzy system[J]. Computerized MedicalImaging and Graphics,2005,29:447-458.
[23] Kawata Y, Niki H N, Ohmatsu, et al. Computerized Analysis of3-D PulmonaryNodule Images In Surrounding and Internal Structure Feature Spaces[J]. IEEEtransactions on medical imaging,2001:889–892.
[24] Penedo M G, Carreira M J, Mosquera A,et al. Computer-Aided Diagnosis:ANeural-Network-Based Approach to Lung Nodule Detection.[J].IEEEtransactions on medical imaging,1998,17(6):872-880.
[25] Suykens J A K, Vandewalle J. Least squares support vector machine classifiers.Neural Processing Letters,1999,9(3),293-300.
[26] Lin Chun-Fu, Wang Sheng-De. Fuzzy support vector machines[C]. IEEETranssaction on Neural Networks,2002,13(2):464-471.
[27]刘露.基于CT图像肺癌定性诊断若干关键问题研究[D].哈尔滨工业大学，2009.
[28]曹蕾.肺结节图像的分析与识别[D].南方医科大学，2009.
[29]李峥嵘，刘晓平，李自元，丁厚本.图像分割多阈值法研究[J]. CT理论与应用研究，2006(04):13-17.
[30]刘东菊.基于阈值的图像分割算法的研究[D].北京交通大学，2009,6.
[31]王亮亮，王黎，高晓蓉，王泽勇.两种改进的局部阈值分割算法[J].现代电子技术，2009(14):78-80.
[32]李越.基于形态学及灰度熵的肺结节检测算法[D].北京交通大学，2011.
[33]王青竹.基于三维SVMs的肺部CT中的结节检测算法[D].吉林大学，2011.
[34]冈萨雷斯.数字图像处理[M].北京：电子工业出版社，2005,9.
[35]徐思瑜.基于CT图像肺部病灶区域的特征提取[D].吉林大学，2011.
[36]何清.模糊聚类分析理论与应用研究进展.模糊系统与数学[J]，1998,12(2):89~94.
[37]高新波，谢维信.模糊聚类理论发展及应用的研究进展[J].科学通报，1999(21):2242-2251.
[38] Zadeh L A. Fuzzy sets[M]. Inf Cont,1965,8:338~353.
[39]焦李成，谭山.图像的多尺度几何分析:回顾和展望[J].电子学报.2003(12):1975-1981.
[40]于立燕.肺部结节的自动检测和特征分析[D].东南大学，2004,1:13-55.
[41]郭亚宁，冯莎莎.机器学习理论研究[J].中国科技信息，010(14):208-214.
[42]覃光华.人工神经网络技术及其应用[D].四川大学，2003.
[43] Suzuki K., Li. F., Sone S., et al. Computer-aided Diagnostic Scheme forDistinction Between Benign and Malignant Nodules in Thoracic Low-dose CT byUse of Massive Training Artificial Neural Network[J]. IEEE Trans. Med. Imaging.2005,24:1138–1150.1.
[44]罗晓华.基于分形的孤立肺结节识别[D].重庆大学，2006.
[45]秦钧.孤立性肺结节的计算机辅助诊断研究[D].重庆大学，2006.
[46]王珏，石纯一.机器学习研究[J].广西师范大学学报(自然科学版)，2003,6(02):1-15.
[47]翟俊海，赵文秀，王熙照.图像特征提取研究[J].河北大学学报(自然科学版).2009,1(01):106-112.
[48]张婧，李彬.结合规则和SVM方法的肺结节识别[J].华南理工大学学报(自然科学版)，2011(02):125-147.
[49] VladimirN. Vapnik统计学习理论的本质[M].北京:清华大学出版社，2000.
[50]张学工.关于统计学习理论与支持向量机[J].自动化学报，2000(01):32-42
[51]王国胜.支持向量机的理论与算法研究[D].北京邮电大学，2008(6):35-49.
[52]张国云.支持向量机算法及其应用研究[D].湖南大学，2006.
[53]阎威武，邵惠鹤.支持向量机和最小二乘支持向量机的比较及应用研究[J].控制与决策.2003,5(03):358-360.
[54]姜静清.最小二乘支持向量机算法及应用研究[D].吉林大学，2007.
[55]李丽娟.最小二乘支持向量机建模及预测控制算法研究[D].浙江大学，2008,6:19-31.
[2]梁琰.肺部CT图像分析及特征提取研究[D].重庆大学，2007.
[3]孙旭辉.基于CT图像的肺结节检测与识别算法研究[D].太原科技大学，2011.
[4]杨合超，周雪梅.几种图像分割技术的比较[J].电脑知识与技术，2009(9):2440-2441.
[5]杨怀义.图像分割中算法的应用研究[J].计算机仿真，2012(02):229-232.
[6]胡敏，李梅，汪荣贵.改进的Otsu算法在图像分割中的应用[J].电子测量与仪器学报，2010(05):443-449.
[7]楚存坤，李月卿，王昌元.医学图像的分割技术及其新进展[N].泰山医学院学报，2007,28(4):315-317.
[8] Samuel G. A., Maryellen L. G., Catherine J. M., et al. Computerized Detection ofPulmonary Nodules on CT Scans. Imaging&Therapeutic Technology[J].1999,19:1303-1311.
[9] Gurcan M. N., Sanhiner B., Petrick N. Lung Nodule Detection on ThoracicComputed Tomography Images: Preliminary Evaluation of a Computer-aidedDiagnosis Systems [J]. Med Phys,2002,29:2552-2558.
[10] M. Dajnowiec, J. Alirezaie. Computer Simulation for Segmentation of LungNodules in CT Images[C]. International Conference on Systems: Man andCybernetics. Toronto,2004:4491~4496.
[11] J.M. Kuhnigk, J. Dicken, L. Bornemann, et al. Morphological Segmentation andPartial Volume Analysis for Volumetry of Solid Pulmonary Lesions in ThoracicCT Scans[C]. IEEE Transactions on Medical Imaging,2006,25(4):417~434.
[12] Lee Y., Hara T., Fujita H., et al. Automated Detection of Pulmonary Nodules inHelical CT Images based on an Improved Template-Matching Technique[J].IEEE Trans. Med. Imaging.2001,20:595–604.
[13] Okumura T, Miwa, Jun-ichi T, Kako, et al. Automatic Detection of Lung CancerChest CT Images by Variable N-Quoit Filter[C].Fourteenth InternationalConference on Pattern Recognition,1998,2(8):1671–1673.
[14] Carol L N, Li F. An interactive system for CT lung nodule identification andexamination[C].Elsevier Science International Congress,2001,1230:639-645.
[15] Li F, Carol L N, Jianzhong Q, et al. Automatic detection of lung nodules frommulti-slice low-dose CT images[J].Proc. SPIE2001,4322:1828-1835.
[16] Y. Yim, H. Hong, Y. G. Shin. Hybrid Lung Segmentation in Chest CT Images forComputer-Aided Diagnosis[C]. Proceedings of7th International Workshop onEnterprise networking and Computing in Healthcare Industry. Kuala,2005:378~383.
[17] Fetita C. I., Prêteux F. J., Beigelman-Aubry C., et al.3-D Automated LungNodule Segmentation in HRCT[J]. Lecture Notes in Computer Science,2003,2878/2003:626–634.
[18]吴龙海.高分辨率CT图像的肺部病变计算机辅助诊断研究[D].中国科学技术大学，2008.
[19]孙申申.基于CT影像的肺结节检测与分割方法研究[D].东北大学，2009.
[20]王倩. CT图像中肺部疾病的计算机辅助诊断方法研究[D].华中科技大学，2009
[21] Suzuki K, Feng Li, Sone S, Doi K. Computer-aided diagnostic scheme fordistinction between benign and malignant nodules in thoracic low-dose CT by useof massive training artificial neural network[C]. IEEE Transactions on MedicalImage,2005,24:1138–1150.
[22] Lin DT, Yan CR, Chen WT. Autonomous detection of pulmonary nodules on CTimages with aneural network-based fuzzy system[J]. Computerized MedicalImaging and Graphics,2005,29:447-458.
[23] Kawata Y, Niki H N, Ohmatsu, et al. Computerized Analysis of3-D PulmonaryNodule Images In Surrounding and Internal Structure Feature Spaces[J]. IEEEtransactions on medical imaging,2001:889–892.
[24] Penedo M G, Carreira M J, Mosquera A,et al. Computer-Aided Diagnosis:ANeural-Network-Based Approach to Lung Nodule Detection.[J].IEEEtransactions on medical imaging,1998,17(6):872-880.
[25] Suykens J A K, Vandewalle J. Least squares support vector machine classifiers.Neural Processing Letters,1999,9(3),293-300.
[26] Lin Chun-Fu, Wang Sheng-De. Fuzzy support vector machines[C]. IEEETranssaction on Neural Networks,2002,13(2):464-471.
[27]刘露.基于CT图像肺癌定性诊断若干关键问题研究[D].哈尔滨工业大学，2009.
[28]曹蕾.肺结节图像的分析与识别[D].南方医科大学，2009.
[29]李峥嵘，刘晓平，李自元，丁厚本.图像分割多阈值法研究[J]. CT理论与应用研究，2006(04):13-17.
[30]刘东菊.基于阈值的图像分割算法的研究[D].北京交通大学，2009,6.
[31]王亮亮，王黎，高晓蓉，王泽勇.两种改进的局部阈值分割算法[J].现代电子技术，2009(14):78-80.
[32]李越.基于形态学及灰度熵的肺结节检测算法[D].北京交通大学，2011.
[33]王青竹.基于三维SVMs的肺部CT中的结节检测算法[D].吉林大学，2011.
[34]冈萨雷斯.数字图像处理[M].北京：电子工业出版社，2005,9.
[35]徐思瑜.基于CT图像肺部病灶区域的特征提取[D].吉林大学，2011.
[36]何清.模糊聚类分析理论与应用研究进展.模糊系统与数学[J]，1998,12(2):89~94.
[37]高新波，谢维信.模糊聚类理论发展及应用的研究进展[J].科学通报，1999(21):2242-2251.
[38] Zadeh L A. Fuzzy sets[M]. Inf Cont,1965,8:338~353.
[39]焦李成，谭山.图像的多尺度几何分析:回顾和展望[J].电子学报.2003(12):1975-1981.
[40]于立燕.肺部结节的自动检测和特征分析[D].东南大学，2004,1:13-55.
[41]郭亚宁，冯莎莎.机器学习理论研究[J].中国科技信息，010(14):208-214.
[42]覃光华.人工神经网络技术及其应用[D].四川大学，2003.
[43] Suzuki K., Li. F., Sone S., et al. Computer-aided Diagnostic Scheme forDistinction Between Benign and Malignant Nodules in Thoracic Low-dose CT byUse of Massive Training Artificial Neural Network[J]. IEEE Trans. Med. Imaging.2005,24:1138–1150.1.
[44]罗晓华.基于分形的孤立肺结节识别[D].重庆大学，2006.
[45]秦钧.孤立性肺结节的计算机辅助诊断研究[D].重庆大学，2006.
[46]王珏，石纯一.机器学习研究[J].广西师范大学学报(自然科学版)，2003,6(02):1-15.
[47]翟俊海，赵文秀，王熙照.图像特征提取研究[J].河北大学学报(自然科学版).2009,1(01):106-112.
[48]张婧，李彬.结合规则和SVM方法的肺结节识别[J].华南理工大学学报(自然科学版)，2011(02):125-147.
[49] VladimirN. Vapnik统计学习理论的本质[M].北京:清华大学出版社，2000.
[50]张学工.关于统计学习理论与支持向量机[J].自动化学报，2000(01):32-42
[51]王国胜.支持向量机的理论与算法研究[D].北京邮电大学，2008(6):35-49.
[52]张国云.支持向量机算法及其应用研究[D].湖南大学，2006.
[53]阎威武，邵惠鹤.支持向量机和最小二乘支持向量机的比较及应用研究[J].控制与决策.2003,5(03):358-360.
[54]姜静清.最小二乘支持向量机算法及应用研究[D].吉林大学，2007.
[55]李丽娟.最小二乘支持向量机建模及预测控制算法研究[D].浙江大学，2008,6:19-31.