Spatio-Temporal Modelling of Perfusion Cardiovascular MRI

Myocardial perfusion MRI provides valuable insight into how
coronary artery and microvascular diseases affect myocardial
tissue. Stenosis in a coronary vessel leads to reduced maximum
blood flow (MBF), but collaterals may secure the blood supply of
the myocardium but with altered tracer kinetics. To date,
quantitative analysis of myocardial perfusion MRI has only been
performed on a local level, largely ignoring the contextual
information inherent in different myocardial segments. This paper
proposes to quantify the spatial dependencies between the local
kinetics via a Hierarchical Bayesian Model (HBM). In the proposed
framework, all local systems are modelled simultaneously along with
their dependencies, thus allowing more robust context-driven
estimation of local kinetics. Detailed validation on both simulated
and patient data is provided.