A Novel Anisotropic Fast Marching Method and its Application to Blood Flow Computation in Phase-contrast MRI

Abstract

SummaryBackground: Phase-contrast MRI (PC MRI) can be used to assess blood flow dynamics noninvasively inside the human body. The acquired images can be reconstructed into flow vector fields. Traditionally, streamlines can be computed based on the vector fields to visualize flow patterns and particle trajectories.Objectives: The traditional methods may give a false impression of precision, as they do not consider the measurement uncertainty in the PC MRI images. In our prior work, we incorporated the uncertainty of the measurement into the computation of particle trajectories.Methods: As a major part of the contribution, a novel numerical scheme for solving the anisotropic Fast Marching problem is presented. A computing time comparison to state-of-theart methods is conducted on artificial tensor fields. A visual comparison of healthy to pathological blood flow patterns is given.Results: The comparison shows that the novel anisotropic Fast Marching solver outperforms previous schemes in terms of computing time. The visual comparison of flow patterns directly visualizes large deviations of pathological flow from healthy flow.Conclusions: The novel anisotropic Fast Marching solver efficiently resolves even strongly anisotropic path costs. The visualization method enables the user to assess the uncertainty of particle trajectories derived from PC MRI images.