Analysis of Cartesian and projection (K,T)-space sampling patterns when using the partially separable functions model for cardiac perfusion MR imaging

Cardiac perfusion MRI aims to analyze perfusion characteristics of the heart through the injection of a contrast agent. In clinical practice, cardiac perfusion imaging is used to diagnose ischemic myocardial tissue, coronary artery disease, and, to a lesser extent, myocardial infarctions. To diagnose these tissues and diseases, real time ECG triggered cardiac imaging techniques (fast T1-weighted gradient echo, echo planar, or steady-state free precession sequences) are typically used to capture the quick wash-in and wash out of the contrast agent. An alternative approach for real-time MRI based on the partially separable functions (PSF) model has been shown to provide reconstructions of dynamic sequences with good spatial and temporal resolutions without the need for ECG triggering. Although previous studies have demonstrated good results using the PSF model, a detailed analysis on the e??ect of the choice of sampling pattern has yet to be performed. Consequently, this thesis aims to analyze the ability of various Cartesian and projection sampling trajectories to characterize perfusion characteristics of the heart when the perfusion is spatially inhomogeneous using the PSF dynamic imaging method. Ten total sampling patterns (five Cartesian and five projection) were analyzed. Overall, the Cartesian sampling patterns provided better reconstructions in terms of image quality, image sequence NRMSE, and relative error of the quantitative perfusion parameters. Also, the Cartesian sampling patterns provided a more accurate reconstruction of the smallest simulated perfusion defect in the myocardium. However, the best sampling pattern performer for the other size defects varied. Also, projection sampling patterns showed a superior robustness to noise compared to Cartesian sampling patterns