High Intraocular Pressure Detection from Frontal Eye Images: A Machine Learning Based Approach.

This paper presents a novel framework to detect the status of intraocular pressure (normal/high) using solely frontal eye image analysis. The framework is based on machine learning approaches to extract six features from frontal eye images. These features include Pupil/Iris ratio, red area percentage, mean redness level of the sclera, and three novel features from the sclera contour (angle, area and distance). Four hundred frontal eye images were used as the image database. The images were taken and annotated by ophthalmologists at Princess Basma Hospital. The proposed framework is fully automated and once the six features were extracted, two classifiers (decision tree and support vector machine) were applied to obtain the status of the eye in terms of eye pressure. The overall accuracy of the proposed framework is 95.5% using the decision tree classifier.