Enhanced Diabetic Retinopathy Detection and Classification Model Using SqueezeNet with Whale Optimization Algorithm

Diabetic Retinopathy (DR) is a severe eye condition predominantly impacting people suffering from diabetes, and it stands as a major contributor to adult blindness. Traditional detection methods involving optometric physicians are laborious and time-consuming. Recent advancements in artificial intelligence (AI) enable more efficient automatic DR detection. This paper presents an Enhanced Diabetic Retinopathy Detection and Classification Model using the Whale Optimization Algorithm (EDRDCM-WOA). The EDRDCM-WOA automates DR classification by employing Gabor filtering (GF) for noise removal during preprocessing. It uses the SqueezeNet method to generate feature vectors and applies the Whale Optimization Algorithm (WOA) for optimal hyperparameter tuning, enhancing classification performance. The Kernel Extreme Learning Machine (KELM) is then used for final classification. Simulations on benchmark dataset demonstrate that the EDRDCM-WOA outperforms recent models, highlighting its effectiveness in DR detection and classification.