A Novel Hybrid Neural Network for Accurate Classification of Retinal Diseases

Retinal diseases like glaucoma, diabetic Retinopathy and cataract result in severe vision impairment or blindness unless they receive treatment early. Early detection is important both for preventing vision deterioration and achieving better outcomes for patients. This research presents a novel hybrid neural network model which achieves efficient and high-performance retinal disease classification. Through this model the extraction of features using CNN becomes possible alongside the ability of RNN to identify sequential patterns present in temporal medical information. This model serves two functions by extracting features from spatial data in retinal images simultaneously with extracting features from temporal patient medical data. This methodology evaluates labeled retinal image data in large databases through its training process to yield superior medical diagnostics above conventional healthcare outputs and demonstrates exceptional potential for retinal disorder diagnosis according to research which enhances diagnostic accuracy and makes it suitable for automated screening along with telemedicine systems.