Nonlinear self-calibrated phase-contrast correction in quantitative cardiac imaging

Abstract

Background Residual background phase in cardiac phase-contrast (PC) imaging introduces velocity errors that bias quantitative flow measurements [1]. While the bias can be offset using static phantoms [2], improved workflow is realized if self-calibrated correction is performed by fitting the phase of static tissue from the in vivo images. However, the residual phase can be nonlinear in space and the vessels of interest, e.g. great vessels, are often far from any static tissue in the image. This means that a linear fit [3] can sometimes result in under-fitting, while fitting with higher spatial-orders can result in over-fitting. Methods We propose a nonlinear self-calibrated approach, which assumes a nonlinear shape. This follows observations that the residual phase is similar in shape to that of the concomitant field. Therefore as compared to linear fitting that uses 4 terms (constant + XYZ), the nonlinear-fit has 5 terms that also include the concomitant field. Further steps are taken to improve the fit, which include iterative removal of outliers that frequently occur at tissue boundaries, and weighting velocities from the quiescent cardiac phase more heavily to reduce effects from flow artifacts at systole. To prevent over-