改进的快速FCM 及SVM 实现糖网白色病灶的自动检测

doi:10.3969/j.issn.0258-8021. 2013. 03.08

摘要
图/表
参考文献
相关文章 (9)

全文: PDF (609 KB) HTML (1 KB)
输出: BibTeX | EndNote (RIS)

摘要为构建基于眼底图像的糖尿病视网膜病变(糖网)自动筛查系统，提出一种基于改进的快速FCM(IFFCM)及SVM的糖网白色病灶自动检测算法。首先，利用改进的快速FCM算法，对彩色眼底图像进行粗分割获取糖网白色病灶候选区域，由于该算法将中值滤波添加到FCM算法的准则函数中，同时利用K-means算法的聚类结果对FCM进行聚类中心初始化，使得该算法克服了传统FCM算法计算复杂度高以及对噪声敏感的缺点；其次，采用两层级联分类结构的SVM对候选区域进行分类，即先利用SVM根据候选区域的特征集将白色病灶提取出来，再利用SVM根据另外的特征集将白色病灶中的硬性渗出与棉绒斑区分开，从而实现眼底图像中糖网白色病灶的自动检测。利用该方法对65幅眼底图像进行糖网白色病灶的自动检测，得到图像水平灵敏度100%，特异性95.0%，准确率98.46%；病灶区域水平（硬性渗出/棉绒斑）灵敏度96.42%/97.15%，阳性预测值90.03%/91.18%；平均一幅图像处理时间35.56 s。结果表明：将改进的快速FCM算法所提供的良好粗分割结果与识别率较高的分类器SVM相结合，使得对糖网白色病灶的自动检测结果较优，即该算法能够高效地自动检测出眼底图像中的糖网白色病灶

	服务

	把本文推荐给朋友
	加入我的书架
	加入引用管理器
	E-mail Alert
	RSS
	作者相关文章
	高玮玮¹ 沈建新^1*王玉亮¹梁春¹左晶^2<

关键词 ：眼底图像, 糖尿病视网膜病变, 模糊C-均值(FCM）, 支持向量机（SVM）, 自动检测

Abstract：To develop an automated diabetic retinopathy (DR) screening system, an automatically detecting approach based on improved and fast FCM (IFFCM) as well as SVM was established and studied. First, color fundus images were segmented by IFFCM, and candidate regions of bright lesions were obtained. The median filter was added to objective function of FCM and the result of Kmeans clustering was used to initialize clustering centers of FCM, so the new algorithm overcome the shortcomings of high complexity and sensitivity to noise. Second, a twolevel SVM classification structure was applied to classify the candidate regions. The bright lesions were picked up by features of candidate regions in stage one. Another group of features were used to discriminate hard exudates from cotton wool spots in stage two; as a result the automated detection of bright lesions in fundus images was accomplished. The approach was tested on a new set of 65 fundus images. With an imagebased criterion, sensitivity of 100%, specificity of 950% and accuracy of 9846% are achieved. Average sensitivity of 9642%/9715% and average positive predict value of 9003%/9118% are also achieved with a lesionbased criterion (hard exudates/cotton wool spots). Furthermore, the average time cost in processing an image is 3556 seconds. Results suggest that the combination of the good result of coarse segmentation provided by IFFCM and higher recognition rate of SVM makes the results of automated detection better. It means that the proposed approach can efficiently detect bright lesions of DR from fundus images.

Key words： fundus images diabetic retinopathy fuzzy C-means（FCM） support vector machine (SVM) automated detection

基金资助:国家高技术研究发展计划(863计划)（2006AA020804）；中央高校基本科研业务费专项（南航NJ20120007）；江苏省普通高校研究生科研创新计划（CXLX11_0218）

引用本文:

高玮玮¹ 沈建新^1*王玉亮¹梁春¹左晶^{2<. 改进的快速FCM 及SVM 实现糖网白色病灶的自动检测[J]. 中国生物医学工程学报, 2013, 32(3): 305-312.

GAO Wei Wei¹SHEN Jian Xin^1*WANG Yu Liang¹LIANG Chun¹ZUO Jing². Automated Detection of Bright Lesions of Diabetic Retinopathy Based on Improved and Fast FCM and SVM. journal1, 2013, 32(3): 305-312.}

链接本文:

http://cjbme.csbme.org/CN/10.3969/j.issn.0258-8021. 2013. 03.08 或 http://cjbme.csbme.org/CN/Y2013/V32/I3/305

［1］Wild S, Roglic G, Green A, et al. Global prevalence of diabetes: estimates for the year 2000 and projections for 2030［J］. Diabetes Care, 2004, 27:1047-1053.
［2］蒲一民, 杨君, 杨田. 糖尿病视网膜病变药物治疗的研究进展［J］. 国际眼科杂志, 2011, 11(12):2134-2137.
［3］Watkins PJ. ABC of diabetic retinopathy［J］. British Medical Journal, 2003, 7(2):105-107.
［4］Sanchez CI, Hornero R, Lopez MI, et al. Retinal image analysis to detect and quantify lesions associated with diabetic retinopathy［C］ // Proceedings of 26th IEEE Annual International Conference on Engineering in Medicine and Biology Society. San Francisco: EMBC, 2004: 1:1624-1627.
［5］Villarroel M, Ciudin A, Hernndez C, et al. Neurodegeneration: An early event of diabetic retinopathy［J］. World J Diabetes, 2010, 1(2): 57-64.
［6］彭金娟, 邹海东, 王伟伟, 等. 上海市北新泾社区糖尿病视网膜病变远程筛查系统的应用研究［J］. 中华眼科杂志, 2010, 46(3): 258-262.
［7］张蕾, 许维强, 谭荣强, 等. 糖尿病视网膜病变筛查方式探讨［J］. 国际眼科杂志, 2010, 10(3): 482-484.
［8］陈喆, 张士胜, 朱惠敏. 糖尿病视网膜病变的国际临床分类分析［J］. 国际眼科杂志, 2011, 11(8): 1394-1401.
［9］Ward NP, Tomlinson S, Taylor CJ. The detection and measurement of exudates associated with diabetic retinopathy［J］. Ophthalmology,1989, 96(1): 80-85.［10］Philips R, Forrester J, Sharp P. Automated detection and quantification of retinal exudates［J］. Graefe’s Arch. Clin. Ophthalmol, 1993, 231(2): 90-94.
［11］Sinthanayothin C, Boyce JF, Williamson TH, et al. Automated detection of diabetic retinopathy on digital fundus images［J］. Diabetic Medicine, 2002, 19:105-112.
［12］Osareh A, Mirmehdi M, Thomas B, et al. Comparative exudate classification using support vector machines and neural networks［C］ // 5th International Conference on Medical Image Computing and ComputerAssisted Intervention. Tokyo: Springer , 2002, 2489: 413-420.
［13］Jaafar HF, Nandi AK, Nuaimy WA. Automated detection of exudates in retinal images using a splitandmerge algorithm［C］ // 18th European Signal Processing Conference.Aalborg: EUSIPCO, 2010: 1622-1626.
［14］Bernhard ME, Ole KH, Ole VL, et al. Screening for diabetic retinopathy using computer based image analysis and statistical classification ［J］. Computer Methods and Programs in Biomedicine, 2000, 62: 165-175.
［15］聂生东, 陈瑛, 顾顺德, 等. 磁共振颅脑图像快速模糊聚类分割算法的研究［J］. 中国生物医学工程学报,2001,20(2):105-109.
［16］Bensaid AM, Hall LO, Bezdek JC, et al. Validityguided(Re) clustering with applicaton to image segmentation［J］. IEEE Trans on Fuzzy System, 1996, 4(2): 112-113.
［17］王卫星, 苏培垠.基于颜色梯度矢量流活动轮廓及支持向量机实现白细胞的提取和分类［J］. 光学精密工程, 2012, 20(12): 2781-2790.
［18］郑一华, 徐立中, 黄风辰. 基于支持向量分类的水质分析应用研究［J］. 仪器仪表学报, 2006, 27(6): 2291-2292.
［19］Snchez CI, Hornero R, López MI, et al. A novel automated image processing algorithm for detection of hard exudates based on retinal images analysis［J］. Med Eng Phys, 2007, 30(3): 350-357.
［20］高玮玮, 沈建新, 王玉亮. 基于数学形态学的快速糖尿病视网膜病变自动检测算法［J］. 光谱学与光谱分析, 2012, 32(4): 760-764.