A Systematic Analysis of Advanced Machine Learning Techniques for Fundus Image-based Diabetic Retinopathy Detection

Diabetic retinopathy (DR) is a leading cause of blindness worldwide, necessitating early detection and accurate classification to mitigate these its progression. Fundus imaging has emerged as a noninvasive and reliable method for Diabetic retinopathy (DR) diagnosis. Recent advancements in machine learning (ML) have significantly improved the precision and efficiency of fundus image-based DR detection. This paper provides a systematic analysis of advanced ML techniques employed in Diabetic retinopathy (DR) classification, emphasizing both traditional and deep learning approaches. It explores preprocessing methods, feature extraction techniques, and state-of-the-art classification algorithms, highlighting their effectiveness and limitations. Key challenges such as imbalanced datasets, variability in image quality, and interpretability are discussed, alongside strategies to address issues. The analysis also examines emerging trends, including hybrid models and explainable AI, offering insights into future research directions. This review aims to serve as a comprehensive resource for researchers and practitioners, guiding the development of more robust and accurate ML-based solutions for DR detection.