Minimizing macrovessel signal in cerebral perfusion imaging using independent component analysis.

The pronounced susceptibility effect of macrovessels in MR bolus-tracking studies induces spots of artificially high blood flow and volume in perfusion parameter images. These high-intensity regions impede the detection of perfusion changes and lead to elevated perfusion parameters in adjacent tissues. The purpose of this work was to explore postprocessing methods to reduce the influence of macrovessel signal in dynamic MRI. After data reduction was performed with the use of a principal component analysis (PCA), an independent component analysis (ICA) was applied to separate signal components of different compartments. Based on this decomposition, the dynamic time series were reconstructed with minimized contributions of macrovessel signal and noise. The influence of the temporal resolution and signal-to-noise ratio (SNR) of the source data were investigated by means of a simulation study. A region-of-interest (ROI)-based analysis of corrected and uncorrected in vivo data demonstrated that the influence of arteries and veins was reduced at least by 50%, while gray matter (GM) and white matter (WM) tissues were nearly unaffected by the correction process. Hemodynamic parameter images of the cerebral blood volume (CBV), cerebral blood flow (CBF), and mean transit time (MTT) were calculated from corrected and uncorrected scans. The corrected parameter images showed a clearly reduced macrovessel signal and an improved perceptibility of microvascular perfusion changes compared to the uncorrected ones.