|Title:||Recognition of hard exudates using Deep Learning|
|Bibliographic citation:||Auccahuasi, W., Flores, E., Sernaque, F., Cueva, J., Diaz, M., Oré, E. (2020). Recognition of hard exudates using Deep Learning. Procedia Computer Science, 167, 2343-2353. https://doi.org/10.1016/j.procs.2020.03.287|
|Abstract:||Diabetes Mellitus is a metabolic disease characterized by the presence of elevated blood glucose levels. Diabetes itself causes other chronic complications, including an eye disease known as diabetic retinopathy. Nowadays, diabetic retinopathy is the most frequent cause of blindness among the active population of developed countries. The principles that produce this disease are not completely known and can not yet be prevented. However, there are effective treatments that delay their evolution as long as it is diagnosed with sufficient anticipation. The problem of diabetic retinopathy is that it is an asymptomatic disease and only defects appear in the vision at an advanced stage of the disease. So in the early stages of diabetic retinopathy is usually imperceptible, diabetic patients do not realize that they have the disease and do not undergo an eye examination. Sometimes the patient is examined when it is too late for proper treatment, due to the presence of severe damage to the retina, occurring only the diagnosis of Diabetes. Currently, technology is becoming more important in the field of health, due to this, a series of systems have been designed to help decision making that helps in the early detection of diabetic retinopathy through the images of Eye, in the present work we present a methodology to be able to recognize the hard exudates that is the first manifestation of diabetic retinopathy, by presenting coloration similar to the other anatomical forms of the eye, its automatic recognition is complicated, the methodology that is presented consists of the use of a database of fundus images with positive and negative symptoms of diabetic retinopathy, from this database a set of images is created that correspond to the hard exudates and images that do not correspond to the hard exudates, with this set of images creates a convolutional network, in order to improve the recognition, obtaining sultados that can satisfy in the clinical practice.|
Repositorio Institucional - Continental
|Appears in Collections:||Artículos Científicos|
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.