Unsupervised Learning for Concept Detection in Medical Images: A Comparative Analysis

As digital medical imaging becomes more prevalent and archives increase in size, representation learning exposes an interesting opportunity for enhanced medical decision support systems. On the other hand, medical imaging data is often scarce and short on annotations. In this paper, we present an assessment of unsupervised feature learning approaches for images in biomedical literature which can be applied to automatic biomedical concept detection. Six unsupervised representation learning methods were built, including traditional bags of visual words, autoencoders, and generative adversarial networks. Each model was trained, and their respective feature spaces evaluated using images from the ImageCLEF 2017 concept detection task. The highest mean F-1 score of 0.108 was obtained using representations from an adversarial autoencoder, which increased to 0.111 when combined with the representations from the sparse denoising autoencoder. We conclude that it is possible to obtain more powerful representations with modern deep learning approaches than with previously popular computer vision methods. The possibility of semi-supervised learning as well as its use in medical information retrieval problems are the next steps to be strongly considered.