Employing deep learning architectures for image-based automatic cataract diagnosis

Various eye diseases affect the quality of human life severely and ultimately may result in complete vision loss. Ocular manifest themselves through mostly visual indicators early or mature stages disease by showing abnormalities optics disc, fovea, other descriptive anatomical structures eye. Cataract is among most harmful that affects millions people leading cause public impairment. It shows major symptoms can be employed for detection before hypermature stage. Automatic diagnosis systems intend to assist ophthalmological experts mitigating burden manual clinical decisions on health care utilization. In this study, a system based color fundus images are addressed cataract disease. Deep learning-based models were performed automatic identification diseases. Two pretrained robust architectures, namely VGGNet DenseNet, detect parts The proposed implemented wide unique dataset includes diverse retinal acquired comparatively low-cost common modality, which considered contribution study. show cataracts different phases represents characteristics cataract. By system, dysfunction associated with could identified achievement compared various traditional up-to-date classification systems. achieves 97.94% rate grading.