Evaluation of Machine Learning Classifiers for Choroiditis Prediction from Fundus Images

Choroiditis is a group of inflammatory eye diseases where inflammation occurs in the eye’s choroid, and Fundus images help in its diagnosis and tracking over time. Inflammatory eye diseases are one of the significant reasons for blindness all over the globe. Hence, it is crucial to develop frameworks that can proficiently diagnose the presence of such diseases. For this work, we collected Fundus images from a tertiary eye hospital and applied machine learning algorithms to classify them into the two classes of normal and choroiditis. At present, we have evaluated four machine-learning techniques on the image data set on basis of accuracy and F1-score. SVM has reported highest classification accuracy of 83% and F1-score of 0.82