MRI is a complex problem from an imaging standardisation perspective – there are many ways of forming images with different types of contrast, and scanners at different sites are often optimised again different sets of requirements – there are a lot of ways to adjust a scanner. This is challenging, particularly for multi-site studies such as clinical trials or population studies is it means that images acquired at different sites have different statistical properties. This is also a challenge for new advanced analysis methods, such as machine learning. We’re investigating new test objects to allow scanner differences to be quantified, collaborating with UK industry and colleagues in clinical science to develop new QA and calibration procedures. We’re also working closely with our colleagues in Data Science to understand and quantify uncertainties in MR measurements, covering the complete imaging pipeline, from table to image formation, to post-processing and inference. We’re also working with our national and international partners to align all this with larger similar efforts elsewhere.