Systematic Literature Review: Detecting Cataract With Deep Learning

Abstract

A cataract is an eye disease that can lead to blindness if left untreated. Many studies have proposed a system for detecting cataracts to reduce the risk of early blindness. To find out what method is used to create a cataract detection system, it is necessary to make a Systematic Literature Review (SLR) on studies that discuss deep learning-based cataract detection. Of each method proposed, the approach to modeling from scratch using the CNN algorithm is the majority of the methods used. In addition, other methods used include Transfer Learning, Fine Tuning Transfer Learning, Final Classifier, and pre-trained models are also the methods used. Each method has its advantages and disadvantages. From the results obtained, the use of a method and its combination must pay attention to many factors such as whether the method used and its combination is by the objectives and problems to be solved, understanding the limitations that can affect the performance of the model must be the concern of researchers to produce an accurate model.