Enhanced vessel detection using edge detection and wavelet fusion for diabetic retinopathy detection

This paper explores an image-based diagnostic model for early detection of diabetic retinopathy using medical image processing techniques. Using a database, which contains images from both healthy individuals and patients with diabetic retinopathy, there search applies a sequence of preprocessing steps including CLAHE, normalization and noise reduction. Two edge detection algorithms, Canny and Frangi, were applied to highlight vascular structures, and the output was used by wavelet-based image fusion to enhance detail clarity. The effectiveness of each method was evaluated using structural similarity metrics and mean percentage error, with results indicating improved vessel detection and diagnostic accuracy when preprocessing was optimized. The study proved robust performance in identifying different features from normal anatomy, suggesting that the integration of advanced image processing with statistical validation can significantly enhance retinal screening. This methodology supports more precise, early-stage clinical assessment and offers promising potential for future diagnostic systems.