DarkNet-19 Based Intelligent Diagnostic System for Ocular Diseases

Untimely detection of ocular diseases is the chief cause of visual impairment among people. Several medical examinations are carried out for diagnosing ophthalmic diseases. Due to more visible symptoms at a later stage, ocular diseases are relatively easier to detect. However, the risk of visual impairment increases with passage of its persistability. Detection of ocular disease at earlier stage helps a lot to eradicate visual disabilities in humans. In order to provide a solution in this scenario, many researchers have used artificial intelligence combined with imaging methods to develop different techniques for earlier detection of ocular diseases. Most of such techniques are not meant for multiple diseases and accuracy of detection is quite low. This study proposed a technique for diagnosing six diseases with an accuracy of 93.6%. Image enhancement technique was applied on fundus images and deep features are extracted to improve the performance to a significant level. The Support Vector Machine classifier with the cubic kernel was used in this investigation and showed area under the curve for receiver operating characteristics of 100%.