Iris imaging for health diagnostics

Abstract

The development of mobile technology and machine learning tools has made it easier than ever to monitor health without visiting a doctor. In this thesis, we explore the use of iris imaging as a medical diagnostic tool. We implement a system in which images captured using a mobile device can be uploaded to and analyzed by a central server. With this platform, we hope to build a large database of standard iris images with labeled medical data and facilitate studies of iris diagnostics. In our implementation, the feature extraction and classification tools built are applied to predict diabetes, through a study conducted in collaboration with researchers at Swami Vivekananda Yoga Anusandhana Samsthana (SVYASA). The results show improvement in prediction accuracy and encourage further development of the server platform for future, large-scale studies.