Design of a model for multistage classification of diabeticretinopathy and glaucoma

Abstract

This  study  addresses  the  escalating  prevalence  of  diabetic  retinopathy  (DR)and  glaucoma,  major  global  causes  of  vision  impairment.    We  propose  aninnovative  iterative  Q-learning  model  that  integrates  with  fuzzy  C-meansclustering to improve diagnostic accuracy and classification speed.  Traditionaldiagnostic frameworks often struggle with accuracy and delay in disease stageclassification,  particularly  in  discerning  complex  features  like  exudates  andveins. Our model overcomes these challenges by combining fuzzy C means withQ learning, enhancing precision in identifying key retinal components. The coreof our approach is a custom-designed 45-layer 2D convolutional neural network(CNN) optimized for nuanced detection of DR and glaucoma stages. Comparedto previous approaches, the performance on the IDRID and SMDG-19 datasetsand associated samples shows a 10.9% rise in precision, an 8.5% improvementin overall accuracy, an 8.3% enhancement in recall, a 10.4% larger area underthe curve (AUC), a 5.9% boost in specificity, and a 2.9% decrease in latency.This  methodology  has  the  potential  to  bring  about  significant  changes  in  thefield of DR and glaucoma diagnosis, leading to prompt medical interventionsand possibly decreasing vision loss.  The use of sophisticated machine learningtechniques in medical imaging establishes a model for future investigations inophthalmology and other clinical situations.