Improved Image Quality when estimating Perfusion Parameters using Bayesian Fitting Algorithm
Irene Klærke Mikkelsen1, Anna Tietze1,2, Lars Ribe1, Anne Obel3, Mikkel Bo Hansen1, and Kim Mouridsen1
1CFIN, Aarhus University, Aarhus, Denmark, 2Dept. of Neuroradiology, Aarhus University Hospital, Aarhus, Denmark, 3Neuroradiology, Aarhus University Hospital, Aarhus, Denmark
Dynamic
Contrast Enhanced Perfusion Imaging (DCE) allows for quantification of the
blood-brain barrier integrity in tumor patients. A key post-processing step is to
fit a hemodynamic model to DCE data. The fitting procedure can, however,
cause spurious voxels and image degradation. We compared the widely used Levenberg-Marquardt
fitting algorithm to a Bayesian algorithm. Image quality was assessed in 42
tumor patients. The Bayesian approach provided the highest image quality scores.
This was confirmed in simulated data with fewer outliers (spurious voxels) when
using the Bayesian approach. The hemodynamic two-compartment model that separates
cerebral blood flow and leakage, provides reliable Ve images, when the robust
Bayesian fitting algorithm is used.
This abstract and the presentation materials are available to members only;
a login is required.
