Noise reduction in three‐dimensional phase‐contrast MR velocity measurementsl

Abstract

The authors have developed a method to reduce noise in three-dimensional (3D) phase-contrast magnetic resonance (MR) velocity measurements by exploiting the property that blood is incompressible and, therefore, the velocity field describing its flow must be divergence-free. The divergence-free condition is incorporated by a projection operation in Hilbert space. The velocity field obtained with 3D phase-contrast MR imaging is projected onto the space of divergence-free velocity fields. The reduction of noise is achieved because the projection operation eliminates the noise component that is not divergence-free. Signal-to-noise ratio (S/N) gains on the order of 15%-25% were observed. The immediate effect of this noise reduction manifests itself in higher-quality phase-contrast MR angiograms. Alternatively, the S/N gain can be traded for a reduction in imaging time and/or improved spatial resolution.

Myocardial velocity gradient imaging by phase contrast MRI with application to regional function in myocardial ischemia

“…In the velocity-encoded phase contrast MRI experiment, the sensitivity to motion and limits in quantification are determined by the strength of the velocity-encoding gradient (35,36). To image directly myocardial motion associated with velocities in the range detected by these phase contrast methods would require an in-plane spatial resolution of ϳ0.1 mm.…”

Variational enhancement and denoising of flow field images

“…The above observations point to a need for physicallymotivated algorithms for flow-field correction, enhancement, and denoising, which combine the adaptability and physical soundness needed to address the noted issues in a satisfactory manner [3,4]. In the present paper, we aim to propose one such algorithm that, together with its generalizations, belongs to the family of variational reconstruction methods.…”

Correction for displacement artifacts in 3D phase contrast imaging

“…Instead of attacking the problem at its source, as proposed here, a few methods have been suggested for improving the accuracy of phase contrast data by applying different fluid dynamic constraints (24,25). The result of these methods may be further improved if they are applied on phase contrast data that first have been corrected for displacement artifacts.…”