Fundeus: Developing Machine Learning Based Interactive System that Diagnoses Cataract, Glaucoma and Diabetic Retinopathy using Fundus Images
Machine learning technology is getting more involved with the health industry in a variety of fields. Ophthalmology, the field of diagnosis and treatment of eye diseases, is a great candidate for machine learning analysis because the algorithms can make better and faster predictions based on image data. In addition to that, data is accumulated throughout the years in the field of Ophthalmology that makes the data difficult to make sense of without applying machine learning.
There are several technologies available to Ophthalmologists; however, most of them are expensive and do not cover the whole diagnosis process. Also, there are few features that may lead to misdiagnosis by the doctors due to the complexity of the image. At that point, machine learning can be a good decision support system for doctors to prevent misdiagnosis.
The Fundeus project aims to diagnose a cataract, glaucoma, and diabetic retinopathy using fundus images by a machine learning-based interactive system. In detail, we proposed a decision support system that is easy to use, benefiting from several machine learning algorithms in the background. Machine learning and deep learning-based algorithms are developed and tested with publicly available datasets such as Kaggle, Messidor, and EyePACS. All in all, the website correctly predicts 8.5 images out of 10.
Project Owners - Salih Berk Dinçer, Tanalp Şengün, Berkay Barlas
Project Advisor - Prof. Dr. Hakan Ürey
#tensorflow #computervision #python #machinelearning #branding #design #health #fundus