Effect of windowing and zero-filled reconstruction of MRI data on spatial resolution and acquisition strategy.

Standard, MR spin-warp sampling strategies acquire data on a rectangular k-space grid. That method samples data from the "corners" of k-space, i.e., data that lie in a region of k-space outside of an ellipse just inscribed in the rectangular boundary. Illustrative calculations demonstrate that the data in the corners of k-space contribute to the useful resolution only if an interpolation method such as a zero-filled reconstruction is used. The consequences of this finding on data acquisition and data windowing strategies are discussed. A further implication of this result is that the spatial resolution of images reconstructed with zero-filling (but without radial windowing) is expected to display angular dependence, even when the phase- and frequency-encoded resolutions are identical. This hypothesis is experimentally verified with a slit geometry phantom. It is also observed that images reconstructed without zero-filling do not display the angular dependence of spatial resolution predicted solely by the maximal k-space extent of the raw data. The implications of these results for 3D contrast-enhanced angiographic acquisitions with elliptical centric view ordering are explored with simulations.