Improving 3D Cine Phase Contrast MRI Aortic Hemodynamics In Vivo Measurements by Means of an Anisotropic Diffusion Filter

3D CINE Phase Contrast Magnetic Resonance Imaging (PCMRI) is considered the technique of election to study in vivo the time varying, complex blood flow structures evolving into arteries [1]. PCMRI allows to obtain a quantitative depiction of the spatial distribution of blood velocities from the acquired phase data and the anatomical image of the district of interest from magnitude data. A major limitation in the application of 3D CINE PCMRI to the clinical practice is the long scan time needed to obtain phase datasets (i.e., blood flow velocities) of sufficient quality for hemodynamic visualizations of time evolving fluid structures or for volumetric flow rates retrospective quantification and analysis. Recently, huge efforts have been done to speed up in vivo acquisitions by implementing/optimizing parallel imaging acquisition schemes as the Sensitivity Encoding (SENSE) [2]. However, the increased reduction factors employed in SENSE scheme to speed up the acquisition lead to acquired PCMRI images affected by marked noise levels, with detrimental effects on measured velocity vector fields. A solution could be offered by the application of an Anisotropic Diffusion Filter (ADF) [3]. Anisotropic filtering is already known for its ability in reducing noise without adding blurring effects, thus preserving fine image details, and it is applied here for the first time to PCMRI data. In this work we propose the application of an ADF strategy to 3D cine PCMRI data of the thoracic aorta obtained with SENSE parallel imaging. The effectiveness of anisotropic filtering in improving image quality for the in vivo hemodynamic characterization of the aorta has been investigated on PCMRI studies acquired with different SENSE reduction factors. The final aim is to denoise and regularize 3D cine velocity maps, preserving, in the meantime, the vessel anatomical contours. This approach could allow to obtain the full 3D characterization of the aortic hemodynamics in acquisition times which are acceptable in the clinical practice.