Enhanced Hemorrhage Detection Based on Deep Reinforcement Hemorrhage Candidate Filtering (DR-HCF) for Diabetic Retinopathy Treatment

Diabetic retinopathy (DR) is a retinal condition caused by diabetes that leads to blindness due to an eye infection. The DR is caused by diabetes mellitus, which affects over 80% of the world's population. Hemorrhage happens when DR is misdiagnosed at an early stage when retinal bleeding has already begun. Early detection of retinal bleeding and infection is crucial for avoiding a potentially blinding scenario. As a result, the clinical analysis of DR uses sophisticated computing approaches to boost the total detection rate. Current study methods need additional time investment to make a correct bleeding diagnosis in retinal analysis. So, in this work, introduce the deep reinforcement hemorrhage candidate filtering (DR-HCF) approach is applied to improve the hemorrhage detection accuracy. Hemorrhage candidate characteristics are derived from retinal pictures using a color and shape analysis and extraction technique. Diabetic retinopathy therapy is administered based on the derived characteristics, reducing the risk of permanent visual loss. Several datasets are used to measure the DR-HCF system's computation time of 24.5%, sensitivity ratio of 98.3%, specificity ratio of 97.6%, and accuracy efficacy ratio of 99.45%.