Semiautomated method for noise reduction and background phase error correction in MR phase velocity data

TLDR

Background phase distortion and random noise degrade the quality of MR phase velocity measurements. The method calculates phase‑velocity image standard deviations over a cardiac cycle, identifies static low‑variance regions to fit and subtract a flat background surface, and removes high‑variance noise by zeroing those regions. The technique was validated on left‑ventricular outflow tract data, producing vector and contour velocity maps overlaid on MR angiograms.

Abstract

Abstract Background phase distortion and random noise can adversely affect the quality of magnetic resonance (MR) phase velocity measurements. A semiauto‐mated method has been developed that substantially reduces both effects. To remove the background phase distortion, the following steps were taken: The time standard deviations of the phase velocity images over a cardiac cycle were calculated. Static regions were identified as those in which the standard deviation was low. A flat surface representing an approximation to the background distortion was fitted to the static regions and subtracted from the phase velocity images to give corrected phase images. Random noise was removed by setting to zero those regions in which the standard deviation was high. The technique is demonstrated with a sample set of data in which the in‐plane velocities have been measured in an imaging section showing the left ventricular outflow tract of a human left ventricle. The results are presented in vector and contour form, superimposed on the conventional MR angiographic images.