浏览全部资源
扫码关注微信
Received：17 August 2017，
Revised：2017-10-19，
Published：16 April 2018
移动端阅览
目的
2
在传统糖尿病视网膜病变（糖网）诊断系统中，微动脉瘤和出血斑病灶检测的精确性决定了最终诊断性能。目前的检测诊断方法为了保证高敏感性而产生了大量假阳性样本，由于数据集没有标注病灶区域导致无法有效地建立监督性分类模型以去除假阳性。为了解决监督性学习在糖网诊断中的问题，提出一种基于多核多示例学习的糖网病变诊断方法。
方法
2
首先，检测疑似的微动脉瘤和出血斑病灶区域，并将其视为多示例学习模型中的示例，而将整幅图像视为示例包，从而将糖网诊断转化为多示例学习问题；其次，提取病灶区域的特征对示例进行描述，并通过极限学习机（ELM）分类算法过滤不相关示例以提升后续多示例学习的分类性能；最后，构建多核图的多示例学习模型对健康图像和糖网病变图像进行分类，以实现糖网病变的诊断。
结果
2
通过对国际公共数据集MESSIDOR进行糖网病变诊断评估实验，获得的准确率为90.1%，敏感性为92.4%，特异性为91.4%，ROC（receiver operating characteristic）曲线下面积为0.932，相比其他算法具有较大性能优势。
结论
2
基于多核多示例学习方法在无需提供病灶标注的情况下，能够高效自动地对糖网病变进行诊断，从而既能避免医学图像中标注病灶的费时费力，又可以免除分类算法中假阳性去除的问题，获得较好的效果。
Objective
2
Diabetic retinopathy (DR) is one of the complications of diabetes and causes severe vision loss and blindness in severe cases if left untreated. A regular eye examination is important for initial diagnosis and early treatment. The change in the blood vessels of the retina is the leading cause of DR. The form of red lesions
such as hemorrhage/microaneurysm (HMA)
is the first explicit sign and an important symptom of DR. Hence
in the traditional DR diagnosis system
the accuracy of HMA lesion detection determines the final diagnosis performance. The diagnosis method produces a large number of false positive samples for high sensitivity
and the supervised classification model is ineffective in removing false positives because the dataset does not label the lesion area. A new algorithm based on multi-kernel and multi-instance learning is proposed to solve the problem of supervised learning in DR diagnosis.
Method
2
First
a multi-scale morphological top-hat transformation is employed to enhance blood vessels and red lesions on the green channel image
then the main vessels of the retina are segmented by thresholding technique on the mask image obtained by binarizing the enhanced image. All regions of interest are generated by subtracting the main vessels from the mask image
and a connected-component labeling technique based on region growing is conducted to detect the suspicious HMA. The detected HMA areas are considered instances
and the entire image is considered a bag. Thus
the problem of DR diagnosis is considered a multi-instance learning problem. Second
a 37D feature based on color
texture
and shape is extracted for each candidate HMA to describe the instance in multi-instance learning. Numerous suspected HMAs are generally obtained to ensure high sensitivity in the initial detection of the lesions
but many HMAs would produce a negative effect on the performance of the multi-instance learning. An extreme learning machine (ELM)-based classifier is accordingly constructed to filter irrelevant instances for improving the multi-instance learning performance. Nevertheless
no such database that contains both the ground truth of DR diagnosis at an image label and HMA segmentation at a lesion label is publicly available. For example
the MESSIDOR dataset contains diagnosis information for DR but not the ground-truth location of HMAs
whereas the E-ophtha dataset contains the location information of HMAs without the diagnosis label. Consequently
the ELM-based classifier trained on the E-ophtha dataset cannot be applied directly to the MESSIDOR dataset due to the difference between datasets. A threshold on the output probability value of the ELM-based classifier is designed to filter the irrelevant instances
and the best threshold can be obtained by cross validation on the training set. Finally
a multi-instance learning method
mi-Graph
which assumes that the instances in a bag are not independently and identically distributed
combined with a multi-kernel learning framework
is adopted for DR diagnosis. The method implicitly constructs graphs by deriving affinity matrices and defines an efficient graph kernel considering clique information. The kernel in the multi-kernel learning is defined as a linear combination of multiple kernels
including Gaussian
polynomial
and linear kernels. As a result
a multi-instance learning model based on multi-kernel graph is constructed to classify the input retinal image into DR or no-DR status.
Result
2
The evaluation is implemented on 1 200 images from the publicly available MESSIDOR dataset
which provides the DR diagnosis results. We verify the effectiveness of the proposed method and the irrelevant instance filtration method. The contributions of different features to DR diagnosis in a multi-kernel learning framework are analyzed. We compare our method with other multi-instance learning methods
such as iterative axis parallel rectangle
expectation-maximization diverse density
citation-k-nearest neighbor
and multiple-instance support vector machine. Our method and other DR diagnosis methods on the MESSIDOR dataset are also compared. Our proposed method achieved an accuracy of 90.1%
sensitivity of 92.4%
specificity of 91.4%
and area under the receiver operating characteristic curve of 0.932. Results show that the proposed method performs better than the other methods and is comparable to previous methods.
Conclusion
2
A multi-instance learning algorithm is introduced into DR diagnosis in this study. The detected HMAs and the entire image are considered instances and a bag of multi-instance learning
respectively. The relationship among the instances in a bag is established by using a kernel graph. A multi-kernel learning framework is adopted to enhance the generalization classification performance. Consequently
a multi-instance learning model based on multi-kernel graph is constructed for DR diagnosis. The experimental results indicate that the proposed approach can be used to diagnose DR efficiently without label information of suspicious lesions to avoid the time-consuming effort of labeling the lesions by specialists and false positive reduction.
LiY C, Shi Y J, Zhang H. The recent advances in the treatment of diabetic heart disease[J]. Medical Recapitulate, 2016, 22(18):3625-3628.
李云翠, 石亚静, 张欢.糖尿病性心脏病的治疗进展[J].医学综述, 2016, 22(18):3625-3628.][DOI:10.3969/j.issn.1006-2084.2016.18.024]
Guo Y, Ma X L. Measurement method of width difference change ratio of arteries and veins in retinal images[J]. Chinese Journal of Scientific Instrument, 2016, 37(4):912-919.
郭莹, 马秀丽.眼底图像动静脉宽度差变化率测量方法[J].仪器仪表学报, 2016, 37(4):912-919.][DOI:10.3969/j.issn.0254-3087.2016.04.025
Zheng S H, Pan L, Chen J, et al. Grading method for non-proliferative diabetic retinopathy images based on microaneurysms and hemorrhages automatic detection[J]. Chinese Journal of Scientific Instrument, 2014, 35(1):59-67.
郑绍华, 潘林, 陈健, 等.微动脉瘤与出血自动检测的NPDR图像分型方法[J].仪器仪表学报, 2014, 35(1):59-67.
Sjølie A K, Stephenson J, Aldington S, et al. Retinopathy and vision loss in insulin-dependent diabetes in Europe:the EURODIAB IDDM complications study[J]. Ophthalmology, 1997, 104(2):252-260.[DOI:10.1016/S0161-6420(97)30327-3]
Akram M U, Khalid S, Khan S A. Identification and classification of microaneurysms for early detection of diabetic retinopathy[J]. Pattern Recognition, 2013, 46(1):107-116.[DOI:10.1016/j.patcog.2012.07.002]
Ren F L, Cao P, Li W, et al. Ensemble based adaptive over-sampling method for imbalanced data learning in computer aided detection of microaneurysm[J]. Computerized Medical Imaging and Graphics, 2017, 55:54-67.[DOI:10.1016/j.compmedimag.2016.07.011]
Oliver A, LladóX, MartíJ, et al. False positive reduction in breast mass detection using two-dimensional PCA[C]//Iberian Conference on Pattern Recognition and Image Analysis. Berlin, Heidelberg: Springer-Verlag, 2007: 154-161. [ DOI:10.1007/978-3-540-72849-8_20 http://dx.doi.org/10.1007/978-3-540-72849-8_20 ]
Boroczky L, Zhao L, Lee K P. Feature subset selection for improving the performance of false positive reduction in lung nodule CAD[J]. IEEE Transactions on Information Technology in Biomedicine, 2006, 10(3):504-511.[DOI:10.1109/TITB.2006.872063]
Cao P, Yang J Z, Li W, et al. Ensemble-based hybrid probabilistic sampling for imbalanced data learning in lung nodule CAD[J]. Computerized Medical Imaging and Graphics, 2014, 38(3):137-150.[DOI:10.1016/j.compmedimag.2013.12.003]
Decencière E, Zhang X W, Cazuguel G, et al. Feedback on a publicly distributed image database:the Messidor database[J]. Image Analysis&Stereology, 2014, 33(3):231-234.[DOI:10.5566/ias.1155]
Li D X, Zhao X Q, Li N. Survey of MIL algorithms for image semantic analysis[J]. Control and Decision, 2013, 28(4):481-488.
李大湘, 赵小强, 李娜.图像语义分析的多示例学习算法综述[J].控制与决策, 2013, 28(4):481-488.
Quellec G, Cazuguel G, Cochener B, et al. Multiple-instance learning for medical image and video analysis[J]. IEEE Reviews in Biomedical Engineering, 2017.[DOI:10.1109/RBME.2017.2651164]
Zhou Z H, Sun Y Y, Li Y F. Multi-instance learning by treating instances as non-I. I. D. samples[C]//Proceedings of the 26th Annual International Conference on Machine Learning. Montreal, Quebec, Canada: IEEE, 2009: 1249-1256. [ DOI:10.1145/1553374.1553534 http://dx.doi.org/10.1145/1553374.1553534 ]
Gönen M, Alpaydin E. Multiple kernel learning algorithms[J]. The Journal of Machine Learning Research, 2011, 12:2211-2268.
Mizutani A, Muramatsu C, Hatanaka Y, et al. Automated microaneurysm detection method based on double ring filter in retinal fundus images[C]//Proceedings of the SPIE 7260, Medical Imaging 2009: Computer-Aided Diagnosis. Florida, United States: SPIE, 2009, 7260: #72601N. [ DOI:10.1117/12.813468 http://dx.doi.org/10.1117/12.813468 ]
Antal B, Hajdu A. An ensemble-based system for microaneurysm detection and diabetic retinopathy grading[J]. IEEE Transactions on Biomedical Engineering, 2012, 59(6):1720-1726.[DOI:10.1109/TBME.2012.2193126]
Zhang B, Zhang L, You J, et al. Microaneurysm (MA) detection via sparse representation classifier with MA and Non-MA dictionary learning[C]//Proceedings of the 20th International Conference on Pattern Recognition. Istanbul: IEEE, 2010: 277-280. [ DOI:10.1109/ICPR.2010.77 http://dx.doi.org/10.1109/ICPR.2010.77 ]
Kandemir M, Hamprecht F A. Computer-aided diagnosis from weak supervision:a benchmarking study[J]. Computerized Medical Imaging and Graphics, 2015, 42:44-50.[DOI:10.1016/j.compmedimag.2014.11.010]
Quellec G, Lamard M, Abràmoff M D, et al. A multiple-instance learning framework for diabetic retinopathy screening[J]. Medical Image Analysis, 2012, 16(6):1228-1240.[DOI:10.1016/j.media.2012.06.003]
Dietterich T G, Lathrop R H, Lozano-Pérez T. Solving the multiple instance problem with axis-parallel rectangles[J]. Artificial Intelligence, 1997, 89(1-2):31-71.[DOI:10.1016/S0004-3702(96)00034-3]
Zhang Q, Goldman S A. EM-DD: an improved multiple-instance learning technique[C]//Proceedings of the 14th International Conference on Neural Information Processing Systems: Natural and Synthetic. Vancouver, British Columbia, Canada: MIT Press, 2001: 1073-1080.
Wang J, Zucker J D. Solving the multiple-instance problem: a lazy learning approach[C]//Proceedings of the Seventeenth International Conference on Machine Learning. San Francisco, CA, USA: Morgan Kaufmann Publishers Inc., 2000: 1119-1126.
Andrews S, Tsochantaridis I, Hofmann T. Support vector machines for multiple-instance learning[J]. Advances in Neural Information Processing Systems, 2003, 15(2):561-568.
Huang G B, Zhu Q Y, Siew C K. Extreme learning machine:theory and applications[J]. Neurocomputing, 2006, 70(1-3):489-501.[DOI:10.1016/j.neucom.2005.12.126]
Ren F L, Li W, Yang J Z, et al. Automatic optic disc localization and segmentation in retinal images by a line operator and level sets[J]. Technology and Health Care, 2016, 24(S2):S767-S776.[DOI:10.3233/THC-161206]
Rakotomamonjy A, Bach F R, Canu S, et al. SimpleMKL[J]. Journal of Machine Learning Research, 2008, 9:2491-2521.
Luenberger D G, Ye Y. Linear and Nonlinear Programming[J]. American Journal of Agricultural Economics, 2015, 67(2):8868-8874.[DOI:10.1016/B0-08-043076-7/00610-0]
相关作者
相关机构
京公网安备11010802024621
