Mapping aortic hemodynamics using 3D cine phase contrast magnetic resonance parallel imaging: evaluation of an anisotropic diffusion filter.

PURPOSE: To propose and evaluate an anisotropic diffusion filter to improve visualization and analysis of the thoracic aorta local hemodynamics from phase-contrast MRI sensitivity encoding imaging.

METHODS: The filter parameters were tailored to the phase-contrast MRI sensitivity encoding data, using a simple calibration procedure. The filter was applied to 20 phase-contrast MR image studies (five subjects acquired with four different sensitivity encoding reduction factors). The filter effect was estimated with respect to image quality (noise in velocity images, σ(n)), regularity of the velocity fields (divergence; relative error in velocity magnitude, and absolute error in flow direction), aorta flow pattern visualization (streamlines, secondary flows) and flow rate quantification.

RESULTS: σ(n) decreased up to three times, divergence, error in velocity magnitude, and absolute error in flow direction decreased (by at least 313, 40, and 10%, respectively), indicating less noisy and more regular velocity fields after filtering. Streamline analysis confirmed the beneficial effect of anisotropic diffusion filter, both visually and quantitatively (streamline numbers increased by 207% in whole cardiac cycle and by 180% in systolic phase). A high correlation (r = 0.99) between the prefiltering and postfiltering aortic flow rate values was found.

CONCLUSION: The anisotropic diffusion filter approach can be considered effective in improving the visualization and analysis of the thoracic aorta hemodynamics from phase-contrast MRI sensitivity encoding images.