PURPOSE: To investigate the limits of phase contrast MRI (PC-MRI)-based measurements of viscous losses and turbulent kinetic energy (TKE) pertaining to spatial resolution, signal-to-noise ratio (SNR), and non-Gaussian intravoxel velocity distributions.
THEORY AND METHODS: High-resolution particle tracking velocimetry data obtained in a realistic aortic phantom with stenotic flow were used to simulate PC-MRI measurements at different resolutions and noise levels. Laminar viscous losses were computed using the spatial gradients of the mean velocity vector field, and TKE levels were derived based on the intravoxel phase dispersion of flow-sensitized PC-MRI measurements.
RESULTS: Increasing the voxel size from 0.625 to 2.5 mm resulted in an underestimation of viscous losses of up to 83%, whereas total TKE values showed errors of <15% and reduced sensitivity to voxel size. Relative errors in viscous loss quantification were found to be less dependent on noise levels when compared with TKE values. In general, a SNR of 20-30 is required for both methods.
CONCLUSION: At spatial resolutions feasible in clinical three-dimensional PC-MRI measurements, viscous losses of stenotic flows are significantly underestimated, whereas TKE shows smaller errors and reduced sensitivity to spatial resolution. Magn Reson Med, 2015. © 2015 Wiley Periodicals, Inc.