Keywords: Cardiovascular: Cardiac, Cardiovascular: Cardiac function, Image acquisition: Machine learning
Cardiac MR (CMR) is a key investigation in cardiovascular medicine. CMR is the best modality for measuring cardiac structure and function, but it is plagued by the need for a clinician to analyse the images. Clinicians are slow, which introduces a bottleneck, and they are often imprecise, which can impact downstream patient care. AI promises to solve these issues, but clinical translation is challenging, as is appropriate evaluation. I will discuss some clinically meaningful measures of AI performance and describe an inline implementation that allows images to be analysed as they are acquired, delivering AI directly to clinical care.