An Optimized Deep Learning based Technique for Grading and Extraction of Diabetic Retinopathy Severities

• DOI: 10.31449/inf.v45i5.3561
• Publication Date: 2021-08-04T22:33:55Z
• Authors: Qiu-ming Zhang; Jing Luo; Korhan Cengiz
• Keywords: Fundus (uterus); Grading (engineering)

The prognosis of Diabetic Retinopathy (DR) requires regular eye examinations, as ophthalmologists depends on fundus segmentation to treat DR pathologies. Automated approaches for detection, and classification have developed an imperative area research the effective diagnosis treatment serious conditions that prevent visual impairment. Diagnosis various lesions, well different severities, helping analyze variations in images take necessary measures before disease progresses. Deep learning techniques evolved a recent advent combat issues conventional machine leaning based methods. An optimized deep framework is proposed this article grading extraction diabetic retinopathy severities. This involves steps like background segmentation, feature set extraction, optimization using Cuckoo search Convolutional Neural Network (CNN) severity grade classification. method was validated two standard datasets MESSIDOR IDRiD. yields accuracy value 97.55%, cross entropy loss 0.367 time intricacy 20 mins 15 secs 98.02% 0.345 22 21 IDRiD dataset; respectively. state-of-the-art comparison depicts CNN provides maximum improvement 10.46% comparative existing methodology. better by procurement investigative outcomes acquired exhibits proficient determination.