Synergic Adversarial Label Learning for Grading Retinal Diseases via Knowledge Distillation and Multi-Task Learning

• Published in: IEEE journal of biomedical and health informatics 2021-10, Vol.25 (10), p.3709-3720
• Main Authors: Ju, Lie, Wang, Xin, Zhao, Xin, Lu, Huimin, Mahapatra, Dwarikanath, Bonnington, Paul, Ge, Zongyuan
• Format: Article
• Language: English
• Subjects: Age; Age related diseases; Algorithms; Classification; Computational modeling; Deep convolutional neural networks; Diabetes; Diabetes mellitus; Diabetic retinopathy; Diabetic Retinopathy - diagnostic imaging; Diseases; Distillation; Exudation; Eye diseases; Fundus Oculi; Hemorrhage; Humans; Image classification; knowledge distillation; Knowledge engineering; Learning; Macular degeneration; Medical imaging; medical imaging classification; Model accuracy; multi-task learning; Object recognition; Physicians; Reliability aspects; Reproducibility of Results; Retina; Retinal Diseases; Retinal images; Retinopathy; Signal monitoring; Task analysis; Training
• ISSN: 2168-2194; 2168-2208
• DOI: 10.1109/JBHI.2021.3052916

The need for comprehensive and automated screening methods for retinal image classification has long been recognized. Well-qualified doctors annotated images are very expensive and only a limited amount of data is available for various retinal diseases such as diabetic retinopathy (DR) and age-related macular degeneration (AMD). Some studies show that some retinal diseases such as DR and AMD share some common features like haemorrhages and exudation but most classification algorithms only train those disease models independently when the only single label for one image is available. Inspired by multi-task learning where additional monitoring signals from various sources is beneficial to train a robust model. We propose a method called synergic adversarial label learning (SALL) which leverages relevant retinal disease labels in both semantic and feature space as additional signals and train the model in a collaborative manner using knowledge distillation. Our experiments on DR and AMD fundus image classification task demonstrate that the proposed method can significantly improve the accuracy of the model for grading diseases by 5.91% and 3.69% respectively. In addition, we conduct additional experiments to show the effectiveness of SALL from the aspects of reliability and interpretability in the context of medical imaging application.