A comparative analysis of pretrained and transfer-learning model for automatic diagnosis of glaucoma

In the recent years, glaucoma is the most generally spotted eye disease in the human that suddenly leads to loss of vision. The glaucoma occurs due to the intraocular pressure in the optic nerve, which averts the transmission of information from the optic nerve to brain. In the current scenario, the medical field has achieved the rapid changes due to the advancement of Artificial Intelligence (AI) technology. Deep Learning (DL) is one of the subfamily of AI, which assists to diagnose the disease in the short interval of time with better accuracy results. The Computer Aided Diagnosis (CAD) is a very effective tool, which helps the physician to diagnose and analyze the disease in an easier manner. In the proposed work, a transfer-learning model designed to diagnose the intra ocular pressure in the optic nerve. This model provides the better validation accuracy of 91.2% with minimized loss function. The training data are obtained from publicly available datasets such as DRIVE, ORIGA and RIM ONE. Using the transfer learning model, the overall training time has been reduced considerably because they are already trained in the millions of images and the inter observability error is minimized.