Continual Self-supervised Learning Considering Medical Domain Knowledge in Chest CT Images

TL;DR

This paper introduces a continual self-supervised learning method for chest CT images that leverages medical domain knowledge, enhancing feature robustness and reducing data interference.

Contribution

It proposes a novel CSSL approach with an enhanced DER, mixup, and feature distillation, specifically tailored for medical imaging to improve continual learning performance.

Findings

Outperforms existing methods on chest CT image tasks

Effectively captures medical domain knowledge in continual learning

Reduces data interference during pretraining

Abstract

We propose a novel continual self-supervised learning method (CSSL) considering medical domain knowledge in chest CT images. Our approach addresses the challenge of sequential learning by effectively capturing the relationship between previously learned knowledge and new information at different stages. By incorporating an enhanced DER into CSSL and maintaining both diversity and representativeness within the rehearsal buffer of DER, the risk of data interference during pretraining is reduced, enabling the model to learn more richer and robust feature representations. In addition, we incorporate a mixup strategy and feature distillation to further enhance the model's ability to learn meaningful representations. We validate our method using chest CT images obtained under two different imaging conditions, demonstrating superior performance compared to state-of-the-art methods.