Automatic Detection of Anatomical Structures in Digital Fundus Retinal Images

This paper proposes a novel system for the automatic detection of important anatomical structures such as the Optic Disc (OD), Blood Vessels and Macula in digital fundus retinal images. The novelty is in extraction of blood vessels and localization of macula. OD localization is done using Principle Component Analysis (PCA) followed by an active contour based approach for accurate segmentation of its boundary. A morphology based approach is proposed for Blood Vessel Detection (BVD). Macula is identified by combining BVD with the property that it is the darkest area in the vicinity of OD. The proposed method is tested on a set of 100 images and the results demonstrate the accuracy of the proposed system. Retinal Image Analysis is a key element in detecting retinopathies in patients. It assists in the automatic detection of pathologies such as diabetic retinopathy (DR), macular degeneration, and glaucoma. Optic Disc (OD), macula and retinal vasculature are all important anatomical structures in the retina. The OD localization and segmentation is a crucial task in an automated retinal image analysis system. It is required as a prerequisite for the detection of exudates and also helps in macula detection, as macula is the darkest area in the neighborhood of OD. Blood Vessel Detection (BVD) is an essential step in medical diagnosis of fundus images as it aids in the diagnosis of ocular diseases. Other applications of retinal vasculature extraction include the treatment of age-related macular degeneration, registration algorithms and personal identification in security applications. Macula is highly sensitive region of the retina responsible for detailed central vision. Macular oedema is a special case of DR caused by the leakage of blood vessels in the macula region. Macular oedema can be treated with laser if detected early enough. Identifying the macula region assumes utmost importance as a first step in the detection of macular oedema.