新疆哈萨克族食管癌图像特征提取及分型方法的探讨
|
摘要
目的利用支持向量机(support vector machine,SVM)对新疆哈萨克族X线食管造影图像进行特征提取及分型研究,为食管癌影像学诊断提供参考。方法随机选取正常食管和蕈伞型食管癌X线造影图像各200幅,运用灰度-梯度共生矩阵法和灰度共生矩阵法提取图像的特征,在SVM类型设置上选择C-SVC,并选择多项式核函数,通过调整C-SVC分类器的参数进行实验。结果共计提取23维特征,利用单一特征算法进行分类,灰度-梯度共生矩阵法分类准确率为72. 75%,灰度共生矩阵法分类准确率为85. 25%,而混合纹理特征的分类准确率为86. 25%。结论将纹理特征与SVM相结合对正常食管与蕈伞型食管癌X线造影图像进行特征提取及分析,具有较高的分类识别率,混合特征把图像纹理和灰度特征有效结合,提高了特征的分类能力,为食管癌的计算机辅助诊断系统的开发奠定了基础。
Objective To study the feature extraction and typing for X-ray images of Xinjiang Kazakh esophageal cancer based on support vector machine( SVM).Methods We randomly selected 200 pieces normal esophagus X-ray image and mushroom esophageal carcinoma X-ray image. Gray gradient co-occurrence matrix and gray level co-occurrence matrix texture features were applied to extract the image features.And the feature classification ability was evaluated by C-SVC classifier.Conducting the experiment repeatedly by adjusting the CSVC classifier. Results Twenty-three features were extracted by gray gradient co-occurrence matrix and gray level co-occurrence matrix. The experimental results showed that using single feature classification, the accuracy rate of gray gradient co-occurrence matrix and gray level co-occurrence matrix classification reached to 72. 75% and 85. 25%,respectively.The accuracy rate of the comprehensive of gray gradient co-occurrence matrix and gray level co-occurrence matrix was 86. 25%. It was more suitable for the classification of normal esophagus and mushroom esophageal carcinoma. Conclusions The characteristics of the normal esophagus and the mushroom esophageal carcinoma are extracted and analyzed with the SVM,which have high classification recognition rate,and laid the foundation for the development of computer-aided diagnosis system for esophageal cancer.
Objective To study the feature extraction and typing for X-ray images of Xinjiang Kazakh esophageal cancer based on support vector machine( SVM).Methods We randomly selected 200 pieces normal esophagus X-ray image and mushroom esophageal carcinoma X-ray image. Gray gradient co-occurrence matrix and gray level co-occurrence matrix texture features were applied to extract the image features.And the feature classification ability was evaluated by C-SVC classifier.Conducting the experiment repeatedly by adjusting the CSVC classifier. Results Twenty-three features were extracted by gray gradient co-occurrence matrix and gray level co-occurrence matrix. The experimental results showed that using single feature classification, the accuracy rate of gray gradient co-occurrence matrix and gray level co-occurrence matrix classification reached to 72. 75% and 85. 25%,respectively.The accuracy rate of the comprehensive of gray gradient co-occurrence matrix and gray level co-occurrence matrix was 86. 25%. It was more suitable for the classification of normal esophagus and mushroom esophageal carcinoma. Conclusions The characteristics of the normal esophagus and the mushroom esophageal carcinoma are extracted and analyzed with the SVM,which have high classification recognition rate,and laid the foundation for the development of computer-aided diagnosis system for esophageal cancer.
引文
[1]刘泓,王海霞.窄带成像放大技术与超声微探头结合在诊断早期食管癌的应用[J].中国药物与临床,2017,17(2):206-207.Liu H,WangHX.Application of narrowband imaging amplification technique combined with ultrasound microprobe in the diagnosis of early esophageal carcinoma[J]. Chinese Remedies&Clinics,2017,17(2):206-207.
[2]张朋,王少康,孙桂菊.食物、营养与食管癌关系的研究进展[J].肿瘤,2011,31(8):768-771.Zhang P,Wang SK,Sun GJ. Research progress in association of food and nutrition with esophageal cancer[J]. Tumor,2011,31(8):768-771.
[3]杨芳,木拉提·哈米提,严传波,等.PCA和SVM在新疆哈萨克族食管癌图像分类中的研究与应用[J].科技通报,2016,32(3):53-57.Yang F,Murat H,Yan CB,etal. Classification of esophageal cancer images for Xinjiang Kazak nationality[J]. Science and Technology Bulletin,2016,32(3):53-57.
[4]郭文佳,贾小莉,王喜艳.哈萨克族食管癌核心家系标本库的建立[J].新疆医科大学学报,2015,38(3):266-268.Guo WJ,Jia XL,Wang Xiyan. The establishment of nuclear family of Kazak’s esophageal cancer[J]. Journal of Xinjiang Medical University,2015,38(3):266-268.
[5] Tixier F,Le Res CC,Hatt M,et al. Intratumor heterogeneity characterized by textural features on baseline 18F-FDG PET images predicts response to concomitantradiochemotherapyin esophagealcancer[J].Journal of Nuclear Medicine,2010,52(3):369-378.
[6]王迪,陈庭毅,陈嘉彬.图像特征提取算法研究[J].电脑知识与技术,2017,15(13):149-150,155.Wang D, Chen TY, Chen JB. Research of image feature extraction algorithm[J]. computer knowledge and technology.2017,15(13):149-150,155.
[7]张学工.关于统计学习理论与支持向量机[J].自动化学报,2000,26(1):32-42.Zhang XG. introduction to statistical learning theory and support vector machines[J].Acta Automatica Sinica,2000,26(1):32-42.
[8]褚洪波.支持向量机理论与算法研究[J].和田师范专科学校学报,2012,31(5):104-106.Zhu HB. Support vector machine theory and algorithm research[J]. Journal of Hotan Teachers College,2012,31(5):104-106.
[9]刘宏兵,柳春华.用于手写数字识别的增量式模糊支持向量机[J].信阳师范学院学报(自然科学版),2014,27(3):421-424.Liu HB,Liu CH. Incremental fuzzy support vector machines for handwritten digit recognition[J]. Journal of Xinyang Normal University(Natural Science Edition),2014,27(3):421-424.
[10]李军,高立群,蔡翔.基于支持向量机的消极性文本识别研究[J].电脑知识与技术,2017,13(6):209-211.Li J,Gao LQ,Cai X. Research on negative text recognition based on support vector machine[J]. Computer Knowledge and Technology,2017,13(6):209-211.
[11]谈蓉蓉.基于支持向量机分类的图像识别研究[J].安徽农业科学,2010,38(26):14756-14757.Tan RR. Image recognition research based on support vector machine classification[J]. Anhui Agricultural Sciences,2010,38(26):14756-14757.
[12]俞春强,付睿,陈艳等.基于信息隐藏的图像感兴趣区域保护算法[J].广西师范大学学报,2016,34(2):67-73.Yu CQ,Fu R,Chen Y,et al. Image region of interest protection algorithm based on information hiding[J]. Journal of Guangxi Normal University,2016,34(2):67-73.
[13]华显立,许贵阳.数字图像中值滤波技术研究[J].电子设计工程,2014,22(11):191-193.Hua XL, Xu GY. Study on digital image median filtering technique[J]. Electronic Design Engineering,2014,22(11):191-193.
[14]吴艳,朱明远,周富强.基于直方图均衡化的渠道冰凌密度计算[J].人民长江,2017,48(14):69-72.Wu Y,Zhu MY,Zhou FQ. Calculation of canal ice density based on histogram equalization[J]. People’s Yangtze River,2017,48(14):69-72.
[15]肖达,王润民,邹孝,等.基于Gabor变换和灰度梯度共生矩阵的超声无损测温研究[J].传感技术学报,2017,30(11):1684-1688.Xiao D,Wang RM,Zou X,et al. A noninvasive temperature measurement based on Gabor transform and gray level gradient co-occurrence matrix using ultrasound[J]. Journal of Sensing Technology,2017,30(11):1684-1688.
[16] Haralick RM. Statistical and structural approaches to texture[J].Proceedings of the IEEE,1979,67(5):786-840.
[17] Haralick RM,Shanmugam K,DisteninI.Textural features for image classification[J]. IEEE Transactions on Systems,Man and Cybernetics,1973,3(6):610-621.
[18]欧国振,孙林慧,薛海双.基于重组超矢量的GMM-SVM说话人辨认系统[J].计算机技术与发展,2017,27(7):51-56.Ou GZ,Sun LH,Xue HS. GMM-SVM speaker identification system with recombination of GMM super vector[J]. Computer Technology and Development,2017,27(7):51-56.
[19] Vapnik VN. The nature of statistical learning theory[M].New York:Springer-Verlag,1995.
[20]田芳,周孝信,于之虹.基于支持向量机综合分类模型和关键样本集的电力系统暂态稳定评估[J].电力系统保护与控制,2017,45(22):1-8.Tian F, Zhou XX, Yu ZH. Power system transient stability assessment based on comprehensive SVM classification model and key sample set[J]. Power System Protection and Control,2017,45(22):1-8.
[21]杜娟,邬麟,陈柏林.食管癌的主要发病因素及其机制[J].中国医药导报,2017,14(7):40-44.Du J,WuL,Chen BL. Inducing factors of esophagus cancer and the underlying mechanisms[J]. China Medical Herald,2017,14(7):40-44.
[2]张朋,王少康,孙桂菊.食物、营养与食管癌关系的研究进展[J].肿瘤,2011,31(8):768-771.Zhang P,Wang SK,Sun GJ. Research progress in association of food and nutrition with esophageal cancer[J]. Tumor,2011,31(8):768-771.
[3]杨芳,木拉提·哈米提,严传波,等.PCA和SVM在新疆哈萨克族食管癌图像分类中的研究与应用[J].科技通报,2016,32(3):53-57.Yang F,Murat H,Yan CB,etal. Classification of esophageal cancer images for Xinjiang Kazak nationality[J]. Science and Technology Bulletin,2016,32(3):53-57.
[4]郭文佳,贾小莉,王喜艳.哈萨克族食管癌核心家系标本库的建立[J].新疆医科大学学报,2015,38(3):266-268.Guo WJ,Jia XL,Wang Xiyan. The establishment of nuclear family of Kazak’s esophageal cancer[J]. Journal of Xinjiang Medical University,2015,38(3):266-268.
[5] Tixier F,Le Res CC,Hatt M,et al. Intratumor heterogeneity characterized by textural features on baseline 18F-FDG PET images predicts response to concomitantradiochemotherapyin esophagealcancer[J].Journal of Nuclear Medicine,2010,52(3):369-378.
[6]王迪,陈庭毅,陈嘉彬.图像特征提取算法研究[J].电脑知识与技术,2017,15(13):149-150,155.Wang D, Chen TY, Chen JB. Research of image feature extraction algorithm[J]. computer knowledge and technology.2017,15(13):149-150,155.
[7]张学工.关于统计学习理论与支持向量机[J].自动化学报,2000,26(1):32-42.Zhang XG. introduction to statistical learning theory and support vector machines[J].Acta Automatica Sinica,2000,26(1):32-42.
[8]褚洪波.支持向量机理论与算法研究[J].和田师范专科学校学报,2012,31(5):104-106.Zhu HB. Support vector machine theory and algorithm research[J]. Journal of Hotan Teachers College,2012,31(5):104-106.
[9]刘宏兵,柳春华.用于手写数字识别的增量式模糊支持向量机[J].信阳师范学院学报(自然科学版),2014,27(3):421-424.Liu HB,Liu CH. Incremental fuzzy support vector machines for handwritten digit recognition[J]. Journal of Xinyang Normal University(Natural Science Edition),2014,27(3):421-424.
[10]李军,高立群,蔡翔.基于支持向量机的消极性文本识别研究[J].电脑知识与技术,2017,13(6):209-211.Li J,Gao LQ,Cai X. Research on negative text recognition based on support vector machine[J]. Computer Knowledge and Technology,2017,13(6):209-211.
[11]谈蓉蓉.基于支持向量机分类的图像识别研究[J].安徽农业科学,2010,38(26):14756-14757.Tan RR. Image recognition research based on support vector machine classification[J]. Anhui Agricultural Sciences,2010,38(26):14756-14757.
[12]俞春强,付睿,陈艳等.基于信息隐藏的图像感兴趣区域保护算法[J].广西师范大学学报,2016,34(2):67-73.Yu CQ,Fu R,Chen Y,et al. Image region of interest protection algorithm based on information hiding[J]. Journal of Guangxi Normal University,2016,34(2):67-73.
[13]华显立,许贵阳.数字图像中值滤波技术研究[J].电子设计工程,2014,22(11):191-193.Hua XL, Xu GY. Study on digital image median filtering technique[J]. Electronic Design Engineering,2014,22(11):191-193.
[14]吴艳,朱明远,周富强.基于直方图均衡化的渠道冰凌密度计算[J].人民长江,2017,48(14):69-72.Wu Y,Zhu MY,Zhou FQ. Calculation of canal ice density based on histogram equalization[J]. People’s Yangtze River,2017,48(14):69-72.
[15]肖达,王润民,邹孝,等.基于Gabor变换和灰度梯度共生矩阵的超声无损测温研究[J].传感技术学报,2017,30(11):1684-1688.Xiao D,Wang RM,Zou X,et al. A noninvasive temperature measurement based on Gabor transform and gray level gradient co-occurrence matrix using ultrasound[J]. Journal of Sensing Technology,2017,30(11):1684-1688.
[16] Haralick RM. Statistical and structural approaches to texture[J].Proceedings of the IEEE,1979,67(5):786-840.
[17] Haralick RM,Shanmugam K,DisteninI.Textural features for image classification[J]. IEEE Transactions on Systems,Man and Cybernetics,1973,3(6):610-621.
[18]欧国振,孙林慧,薛海双.基于重组超矢量的GMM-SVM说话人辨认系统[J].计算机技术与发展,2017,27(7):51-56.Ou GZ,Sun LH,Xue HS. GMM-SVM speaker identification system with recombination of GMM super vector[J]. Computer Technology and Development,2017,27(7):51-56.
[19] Vapnik VN. The nature of statistical learning theory[M].New York:Springer-Verlag,1995.
[20]田芳,周孝信,于之虹.基于支持向量机综合分类模型和关键样本集的电力系统暂态稳定评估[J].电力系统保护与控制,2017,45(22):1-8.Tian F, Zhou XX, Yu ZH. Power system transient stability assessment based on comprehensive SVM classification model and key sample set[J]. Power System Protection and Control,2017,45(22):1-8.
[21]杜娟,邬麟,陈柏林.食管癌的主要发病因素及其机制[J].中国医药导报,2017,14(7):40-44.Du J,WuL,Chen BL. Inducing factors of esophagus cancer and the underlying mechanisms[J]. China Medical Herald,2017,14(7):40-44.
© 2004-2018 中国地质图书馆版权所有 京ICP备05064691号 京公网安备11010802017129号 地址：北京市海淀区学院路29号 邮编：100083 电话：办公室：(+86 10)66554848；文献借阅、咨询服务、科技查新：66554700 |