Automatic Cataract Classification Based on Multi-feature Fusion and SVM

According to the World Health Organization (WHO), cataract is the most common cause of vision loss and blindness. In this paper, we propose a scheme to grade cataract into six classifications for precise automatic cataract diagnosis. We extract two kinds of features: texture features by gray-level co-occurrence matrix (GLCM) and high-level features via the pre-trained residual network (ResNet). Then, the two kinds of features are fused in a way of dimension expansion, which texture feature vectors are added to the tail of high-level feature vectors. Next, the fused feature vectors are put into support vector machine (SVM) to train and verify. Our scheme can achieve 91.5% average accuracy on cataract for six classifications. In addition, our proposed scheme outperforms the existing methods for four classifications significantly.

[1]  Max Q.-H. Meng,et al.  Texture analysis for ulcer detection in capsule endoscopy images , 2009, Image Vis. Comput..

[2]  Max A. Viergever,et al.  Ridge-based vessel segmentation in color images of the retina , 2004, IEEE Transactions on Medical Imaging.

[3]  Bram van Ginneken,et al.  Comparative study of retinal vessel segmentation methods on a new publicly available database , 2004, SPIE Medical Imaging.

[4]  Qinyan Zhang,et al.  Classification of cataract fundus image based on deep learning , 2017, 2017 IEEE International Conference on Imaging Systems and Techniques (IST).

[5]  Ana Maria Mendonça,et al.  Segmentation of retinal blood vessels by combining the detection of centerlines and morphological reconstruction , 2006, IEEE Transactions on Medical Imaging.

[6]  Vladimir Vapnik,et al.  An overview of statistical learning theory , 1999, IEEE Trans. Neural Networks.

[7]  F. T. Haar.  Automatic localization of the optic disc in digital colour images of the human retina , 2005 .

[8]  Jian Sun,et al.  Deep Residual Learning for Image Recognition , 2015, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[9]  Xiaoyi Jiang,et al.  Adaptive Local Thresholding by Verification-Based Multithreshold Probing with Application to Vessel Detection in Retinal Images , 2003, IEEE Trans. Pattern Anal. Mach. Intell..

[10]  Qiang Yang,et al.  A Survey on Transfer Learning , 2010, IEEE Transactions on Knowledge and Data Engineering.

[11]  Qinyan Zhang,et al.  Application of SVM based on genetic algorithm in classification of cataract fundus images , 2017, 2017 IEEE International Conference on Imaging Systems and Techniques (IST).

[12]  A.D. Hoover,et al.  Locating blood vessels in retinal images by piecewise threshold probing of a matched filter response , 2000, IEEE Transactions on Medical Imaging.