Self-supervised Mean Teacher for Semi-supervised Chest X-ray Classification
This addresses the high cost of expert annotations in medical imaging by enabling better model training with fewer labeled images, though it appears incremental as it builds on existing semi-supervised and self-supervised techniques.
The paper tackles the problem of limited annotated data in medical image analysis by proposing Self-supervised Mean Teacher for Semi-supervised (S^2MTS^2) learning, which combines self-supervised pre-training with semi-supervised fine-tuning and outperforms previous state-of-the-art methods by a large margin on datasets like Chest X-ray14 and CheXpert.
The training of deep learning models generally requires a large amount of annotated data for effective convergence and generalisation. However, obtaining high-quality annotations is a laboursome and expensive process due to the need of expert radiologists for the labelling task. The study of semi-supervised learning in medical image analysis is then of crucial importance given that it is much less expensive to obtain unlabelled images than to acquire images labelled by expert radiologists. Essentially, semi-supervised methods leverage large sets of unlabelled data to enable better training convergence and generalisation than using only the small set of labelled images. In this paper, we propose Self-supervised Mean Teacher for Semi-supervised (S$^2$MTS$^2$) learning that combines self-supervised mean-teacher pre-training with semi-supervised fine-tuning. The main innovation of S$^2$MTS$^2$ is the self-supervised mean-teacher pre-training based on the joint contrastive learning, which uses an infinite number of pairs of positive query and key features to improve the mean-teacher representation. The model is then fine-tuned using the exponential moving average teacher framework trained with semi-supervised learning. We validate S$^2$MTS$^2$ on the multi-label classification problems from Chest X-ray14 and CheXpert, and the multi-class classification from ISIC2018, where we show that it outperforms the previous SOTA semi-supervised learning methods by a large margin.