A Comprehensive Study of Supervised Learning Models on the OCTID Dataset for Retinal Disease Detection with Future Directions

Abstract

This paper presents a critical analysis of recent approaches to supervised learning techniques on the OCTID database that are important in the early diagnosis of retinal diseases such as AMD, CSR, DME, and MH. The review concentrates on methodologies, experimental configurations, and assessments of various machine learning and deep learning techniques. The limitations observed in the current work are in terms of computational complexity, possible limitation in labeling, and lack of extensive experimentation using videos with OCT. For future work, we suggest using semi-supervised learning as they are more accurate and less costly in medical applications.