The development, deployment, and evolution of representation learning has been used in radiology for intelligent diagnosis, treatment outcome prediction, and biomarker discovery. Representation learning explores how to transform data into quantitative features and to facilitate automatic data analysis.

At present, radiomics and deep learning are still in development, and challenges still exist – e.g., how to automatically extract features with clinical meanings, how to train a deep network with a small number of data samples, how to fuse multi-source information, and how to design representation learning with high interpretability.

This Special Issue calls for submissions of original research and review articles to address these challenges and to highlight the recent progress of representation learning in radiology and related fields. We are particularly interested in articles that could deepen our understanding of representation learning in clinical applications with high interpretability. In addition, articles to uncover clinical and technical challenges are also welcomed.

Potential topics include but are not limited to the following:

• Biomedical data representation and automatic data analysis
• Recent progress in radiomics, delta radiomics and deep learning
• Advanced technologies in multi-source information fusion
• Feature engineering in computer-aided detection and diagnosis
• Representation learning for disease diagnosis and biomarker discovery
• Data representation in the prediction of treatment outcome
• Integrated studies of representation learning and clinical applications