Glaucoma detection in retinal fundus images using U-Net and supervised machine learning algorithms

Glaucoma is a neuro-degenerative eye disease developed due to an increase in the Intra-ocular Pressure inside the retina. Being the second largest cause of blindness worldwide, it can lead the person towards complete blindness if an early diagnosis does not take place. With respect to this underlying issue, there is an immense need of developing a system that can effectively work in the absence of excessive equipments, skilled medical practitioners and also is less time consuming. This work proposes an offline Computer-Aided Diagnosis (CAD) system for glaucoma diagnosis using retinal fundus images. This application is developed using image processing, deep learning and machine learning approaches. Le-Net architecture is used for input image validation and Region of Interest (ROI) detection is done using brightest spot algorithm. Further, the optic disc and optic cup segmentation is performed with the help of U-Net architecture and classification is done using SVM, Neural Network and Adaboost classifiers. The accuracy of 99% is achieved using Le-Net for input image validation. Considering the application of brightest spot algorithm, an accuracy of 98.67% is achieved for ROI extraction. Further, a dice-coefficient of 0.93 and 0.87 was attained for the segmentation of optic disc and optic cup respectively using U-Net architecture. Using SVM, Neural Network and Adaboost classifiers, the proposed methodology managed to achieve a classification accuracy, recall, specificity and sensitivity of 100%, thus proving the system to be reliable and promising. In conclusion, the proposed desktop application is easy to use and can play a major role in early glaucoma detection. The modular design of the CAD system is made up of a set of standalone components that can be used for a variety of different tasks for detecting and classifying glaucoma. Due to the model being trained on a variety of different datasets, the system proves to be robust and more accurate.