Weakly-supervised Generative Adversarial Networks for medical image classification

TL;DR

This paper introduces WSGAN, a novel weakly-supervised GAN-based method for medical image classification that leverages unlabeled data, contrastive learning, and cyclic consistency to improve accuracy with limited labeled samples.

Contribution

The paper proposes WSGAN, integrating MixMatch, contrastive learning, and cyclic consistency to enhance medical image classification with minimal labeled data.

Findings

WSGAN achieves 11% higher accuracy than MixMatch with limited labeled data.

The method effectively utilizes unlabeled data to improve classification performance.

Ablation studies confirm the effectiveness of each component in WSGAN.

Abstract

Weakly-supervised learning has become a popular technology in recent years. In this paper, we propose a novel medical image classification algorithm, called Weakly-Supervised Generative Adversarial Networks (WSGAN), which only uses a small number of real images without labels to generate fake images or mask images to enlarge the sample size of the training set. First, we combine with MixMatch to generate pseudo labels for the fake images and unlabeled images to do the classification. Second, contrastive learning and self-attention mechanism are introduced into the proposed problem to enhance the classification accuracy. Third, the problem of mode collapse is well addressed by cyclic consistency loss. Finally, we design global and local classifiers to complement each other with the key information needed for classification. The experimental results on four medical image datasets show that WSGAN can obtain relatively high learning performance by using few labeled and unlabeled data. For example, the classification accuracy of WSGAN is 11% higher than that of the second-ranked MIXMATCH with 100 labeled images and 1000 unlabeled images on the OCT dataset. In addition, we also conduct ablation experiments to verify the effectiveness of our algorithm.