Integration of Self-Supervised BYOL in Semi-Supervised Medical Image Recognition

TL;DR

The paper addresses the scarcity of labeled medical images by integrating self-supervised learning via BYOL with semi-supervised learning. It pre-trains on unlabeled data using BYOL and then fine-tunes with a mix of labeled and pseudo-labeled samples to construct a classifier. Experiments on OCT2017, COVID-19 X-ray, and Kvasir show higher accuracy than several baselines, with reported scores such as 0.966, 0.987, and 0.976, respectively. This approach reduces labeling requirements while delivering robust medical image recognition performance in data-scarce settings.

Abstract

Image recognition techniques heavily rely on abundant labeled data, particularly in medical contexts. Addressing the challenges associated with obtaining labeled data has led to the prominence of self-supervised learning and semi-supervised learning, especially in scenarios with limited annotated data. In this paper, we proposed an innovative approach by integrating self-supervised learning into semi-supervised models to enhance medical image recognition. Our methodology commences with pre-training on unlabeled data utilizing the BYOL method. Subsequently, we merge pseudo-labeled and labeled datasets to construct a neural network classifier, refining it through iterative fine-tuning. Experimental results on three different datasets demonstrate that our approach optimally leverages unlabeled data, outperforming existing methods in terms of accuracy for medical image recognition.

Figures (2)

• Figure 1: BYOL architecture.
• Figure 2: The proposed semi-supervised learning with BYOL.