Accurate detection of blood vessels improves the detection of exudates in color fundus images

Exudates are one of the earliest and most prevalent symptoms of diseases leading to blindness such as diabetic retinopathy and macular degeneration. Certain areas of the retina with such conditions are to be photocoagulated by laser to stop the disease progress and prevent blindness. Outlining these areas is dependent on outlining the lesions and the anatomic structures of the retina. In this paper, we provide a new method for the detection of blood vessels that improves the detection of exudates in fundus photographs. The method starts with an edge detection algorithm which results in a over segmented image. Then the new feature-based algorithm can be used to accurately detect the blood vessels. This algorithm considers the characteristics of a retinal blood vessel such as its width range, intensities and orientations for the purpose of selective segmentation. Because of its bulb shape and its color similarity with exudates, the optic disc can be detected using the common Hough transform technique. The extracted blood vessel tree and optic disc could be subtracted from the over segmented image to get an initial estimate of exudates. The final estimation of exudates can then be obtained by morphological reconstruction based on the appearance of exudates. This method is shown to be promising since it increases the sensitivity and specificity of exudates detection to 80% and 100% respectively.

[1]  J. Olson,et al.  A comparative evaluation of digital imaging, retinal photography and optometrist examination in screening for diabetic retinopathy , 2003, Diabetic medicine : a journal of the British Diabetic Association.

[2]  B. Thomas,et al.  Automated identification of diabetic retinal exudates in digital colour images , 2003, The British journal of ophthalmology.

[3]  Mong-Li Lee,et al.  An effective approach to detect lesions in color retinal images , 2000, Proceedings IEEE Conference on Computer Vision and Pattern Recognition. CVPR 2000 (Cat. No.PR00662).

[4]  Mong-Li Lee,et al.  The role of domain knowledge in the detection of retinal hard exudates , 2001, Proceedings of the 2001 IEEE Computer Society Conference on Computer Vision and Pattern Recognition. CVPR 2001.

[5]  Hideki Kuga,et al.  A computer method of understanding ocular fundus images , 1982, Pattern Recognit..

[6]  Yasser M. Kadah,et al.  Survey of Retinal Image Segmentation and Registration , 2006 .

[7]  T. Sano,et al.  [Diabetic retinopathy]. , 2001, Nihon rinsho. Japanese journal of clinical medicine.

[8]  Michael H. Goldbaum,et al.  Locating the optic nerve in a retinal image using the fuzzy convergence of the blood vessels , 2003, IEEE Transactions on Medical Imaging.

[9]  Suman K. Mitra,et al.  A Bayesian network based sequential inference for diagnosis of diseases from retinal images , 2005, Pattern Recognit. Lett..

[10]  Yasser M. Kadah,et al.  A new real-time retinal tracking system for image-guided laser treatment , 2002, IEEE Transactions on Biomedical Engineering.

[11]  Jacob Scharcanski,et al.  A coarse-to-fine strategy for automatically detecting exudates in color eye fundus images , 2010, Comput. Medical Imaging Graph..

[12]  Langis Gagnon,et al.  Procedure to detect anatomical structures in optical fundus images , 2001, SPIE Medical Imaging.

[13]  Pierre Soille,et al.  Morphological Image Analysis: Principles and Applications , 2003 .

[14]  Bunyarit Uyyanonvara,et al.  Automatic detection of diabetic retinopathy exudates from non-dilated retinal images using mathematical morphology methods , 2008, Comput. Medical Imaging Graph..