The relationship between spatial resolution levels and quantitative myocardial perfusion

Background Dynamic contrast enhanced cardiovascular magnetic resonance imaging (DCE-CMR) allows for high-resolution analysis of myocardial perfusion by preserving important spatial information. However high resolution perfusion quantification suffers from the poor signal-to-noise ratio (SNR) of voxel-based data. Previously spatial averaging (segmental or sub-segmental analysis) has been used to improve the SNR. This approach results in reduction of spatial resolution and potentially in a loss of information about the extent, localization and transmurality of ischemia. This study investigates the relationship between the level of spatial resolution and accuracy of perfusion estimates obtained using different deconvolution methods including Fermi function modelling, ARMA, B-spline basis and exponential basis deconvolution. A new quality of fit analysis method which uses autocorrelation function properties [1] is used here to measure the fraction of remaining un-modelled information (FRI) of the data (i.e. the amount of diagnostic information lost).