An Approach for the Early Detection of Retinal Disorders and Performing Human Authentication
| Autor: | Chandrashekar M Patil, G. R. Prashantha |
|---|---|
| Rok vydání: | 2017 |
| Předmět: |
Retina
medicine.medical_specialty Retinal Disorder business.industry Insulin medicine.medical_treatment Image processing Diabetic retinopathy medicine.disease medicine.anatomical_structure Diabetes mellitus Ophthalmology medicine Maculopathy Computer vision Artificial intelligence business Blood vessel |
| Zdroj: | Proceedings of International Conference on Cognition and Recognition ISBN: 9789811051456 |
| DOI: | 10.1007/978-981-10-5146-3_16 |
| Popis: | Diabetes is a serious disease which is caused due to the high blood sugar level or in other words due to the reduced insulin production in the body. Prolonged diabetes affects the blood vessels present in the eye and is termed as Diabetic Retinopathy. Diabetic Maculopathy is one such disease suffered by retinopathic patients which results when the fluid leaks out from the blood vessels that are damaged and gather near the central region of retina called Macula. In this paper an approach to detect abnormalities such as blood vessels, micro aneurysms exudates using image processing techniques in the fundus image. These features are used for the detection of severity of Diabetic Retinopathy, Diabetic maculopathy. Authentication of a person can be done based on matching the extracted vessel pattern of the retina with the reference. The algorithm so presented detects the Diabetic Retinopathy and classifies it according to its severity levels. It also detects Maculopathy at the early stage of the disease and performs the authentication of a person based on the blood vessel pattern matching. This system intend to help ophthalmologist in the screening process to detect symptoms of diabetic retinopathy, diabetic maculopathy quickly and more easily. The proposed algorithm is tested over 4 different databases. The multiclass SVM classifies the input retinal image into different classes of disorder as severe, mild, moderate diabetic retinopathy, healthy and maculopathy. The classification is done based on the color matching and SVM classifier by calculating the average intensity, variance, standard deviation, median and centroid. The algorithm is tested over the readily available DRIVE, HRF, DIARETDB0 and DIARETDB1 databases. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|