Cardiovascular Diseases (CVDs) cause more than 26% of all deaths in the UK, costing over £15 billion each year. There is good evidence that a large array of CVDs can be diagnosed by an assessment of heart motion abnormalities. The motion of the heart is acquired using a cine CMR imaging technique, generating a sequence of images across a cardiac cycle at various slices through the heart.
In this project, we will propose a scalable probabilistic approach for holistic motion atlas modeling from a big population data (UK Biobank Cardiac Imaging) comprising >= 20k patients. The motion will be modelled as the spatiotemporal (3D+t) sequence of the heart shapes across a full cardiac cycle, extracted from cine CMR images.
The atlas will be a recurrent deep model that, given a cine sequence, will predict a probabilistic distribution function (pdf) for the next status of the heart. More importantly, the pdf will be conditional on the patient metadata (genomics, age, gender, lifestyle, etc.). Thus, by measuring the spatial deviations from the expected shape at each phase, it will allow accurate and personalized quantification of functional abnormality maps.