Evaluation of an AIF correction algorithm for dynamic susceptibility contrast‐enhanced perfusion MRI

Abstract

For longitudinal studies in patients suffering from cerebrovascular diseases the poor reproducibility of perfusion measurements via dynamic susceptibility-weighted contrast-enhanced MRI (DSC-MRI) is a relevant concern. We evaluate a novel algorithm capable of overcoming limitations in DSC-MRI caused by partial volume and saturation issues in the arterial input function (AIF) by a blood flow stimulation-study. In 21 subjects, perfusion parameters before and after administration of blood flow stimulating L-arginine were calculated utilizing a block-circulant singular value decomposition (cSVD). A total of two different raters and three different rater conditions were employed to select AIFs: Besides 1) an AIF selection by an experienced rater, a beginner rater applied a steady state-oriented strategy, returning; 2) raw; and 3) corrected AIFs. Highly significant changes in regional cerebral blood flow (rCBF) by 9.0% (P < 0.01) could only be found when the AIF correction was performed. To further test for improved reproducibility, in a subgroup of seven subjects the baseline measurement was repeated 6 weeks after the first examination. In this group as well, using the correction algorithm decreased the SD of the difference between the two baseline measurements by 42%. Magn Reson Med 60:102-110, 2008.

Determination of the venous output function from MR signal phase: Feasibility for quantitative DCE-MRI in human brain

“…1. For each VOF and AIF, the AUC of the first pass was calculated using a γ‐variate curve fit to reduce the effects of recirculation (4, 15, 20, 30, 36, 38).…”

Absolute quantification of perfusion using dynamic susceptibility contrast MRI: pitfalls and possibilities

“…The algorithm included fitting a gamma-variate function to the measured concentration-time curve of the artery, modified to include only the points on the signal curve that did not fall below a specific threshold. Brunecker et al [64,65] proposed and evaluated an algorithm for correcting distortions of the AIF caused by peak saturation of the concentration curve. This algorithm separated the AIF into a reliable part at low contrast agent concentrations and an unreliable part at high concentrations.…”

The 39 steps: evading error and deciphering the secrets for accurate dynamic susceptibility contrast MRI

“…In a different approach, Brunecker et al . proposed a post‐processing method to correct distorted AIFs by separating the profile into ‘reliable’ and ‘unreliable’ phases, using a parametric description to reconstruct the most probable profile of the unreliable phase. However, like any parametric model approach (see step 23), the accuracy of this correction method will depend on the validity of the model used.…”