RESEARCH THEME
Uncertainty
Enabling AI models to express their confidence or uncertainty in outputs is crucial when considering AI as a tool to support human decision making.
OUR WORK IN UNCERTAINTY
Uncertainty-aware training
AI models based upon deep learning are notoriously bad at quantifying the confidence of their outputs. By training models using 'hard' labels, solely to maximise performance (e.g. in terms of classification accuracy), they tend to over-estimate confidence. In this case we say that the models are poorly 'calibrated' with respect to uncertainty. We have pioneered the development of training methods that not only aim to maximise accuracy but also calibration, which we refer to as 'uncertainty-aware training'. By demonstrating these methods on real medical imaging problems we hope to advance the adoption of AI models for decision support in medicine.
Publications:
- "Uncertainty Aware Training to Improve Deep Learning Model Calibration for Classification of Cardiac MR Images" Medical Image Analysis, 80, 2023. paper
- "Addressing Deep Learning Model Calibration Using Evidential Neural Networks and Uncertainty-Aware Training" Proceedings ISBI, 2023. paper
Uncertainty in Biomarker Estimation
We have also investigated uncertainty in the context of improving the precision of cardiac MR quantification. We developed a cardiac MR segmentation pipeline that estimated functional biomarkers together with confidence limits, and then defined new metrics for precision quantification that accounted for this uncertainty.
Publications:
- "Confidence Matters: Uncertainty Quantification and Precision Assessment of Deep Learning-based CMR Biomarker Estimates Using Scan-rescan Data" Proceedings ISBI, 2026. paper
Interested in our work?
Explore our research or get in touch with the group.
Contact the group →