Optimizing Microaneurysm Detection for Early Diagnosis of Diabetic Retinopathy with Enhanced Fundus Images

Abstract

Diabetic Retinopathy (DR) stands as the primary cause of visual impairment among individuals in the workforce. Detecting it early through routine screenings and providing accurate treatment has proven effective in preventing vision loss. Fundus photography enables the capture of detailed images of the eye's interior. The automated identification of lesions offers a straightforward and economical approach. This technology aids doctors in decreasing their workload while enhancing accuracy and efficiency. The focus of our research revolves around identifying red lesions, characterized by minuscule dots in fundus images. These dots signify the presence of microaneurysms, which represent an initial stage in diabetic retinopathy detection. Given that microaneurysms serve as symptomatic indicators of diabetic retinopathy, their identification holds crucial importance. Our approach is built upon a highly efficient level set framework, designed to effectively detect the presence of these lesions within the affected images.