In-vivo quantitative assessment of perfusion MRI in humans and animals

Dynamic contrast or the bolus chasing technique, based on magnetic resonance imaging (MRI), allows quantitative and noninvasive evaluation of tissue perfusion. Dynamic contrast magnetic resonance imaging (DC-MRI) involves the administration of a paramagnetic contrast agent which induces local magnetic field gradients as it passes through the vasculature resulting in signal loss. The time-signal intensity changes can be analyzed for quantitative estimation of the hemodynamic parameters such as cerebral blood flow (CBF), cerebral blood volume (CBV), and mean transit time (MTT) that characterize tissue perfusion.;In spite of the wide use of MRI-based perfusion imaging, there are a number of technical problems that need to be addressed and resolved for perfusion imaging to become a robust and routine technique. At present, these include (1) automatic determination of the arterial input function (AIF), (2) using an appropriate deconvolution algorithm to calculate hemodynamic parameters reproducibly and accurately, (3) correcting the bias introduced by large vessel structures by registering perfusion images with susceptibility-weighted images (SWI), and (4) development of a user-friendly software package to process the perfusion data.;In this study, an integrated approach has been implemented for in-vivo brain perfusion studies. An automatic AIF identification algorithm was developed to identify AIF without human intervention. A singular value decomposition deconvolution algorithm, in which the thresholds are automatically determined on a pixel-by-pixel basis, was developed to calculate hemodynamic parameters. SWI images were registered with perfusion images to remove the bias introduced by the large vessels in the estimation of hemodynamic parameters. Using these techniques, perfusion images were processed with an advanced integrated perfusion image processing system developed as a part of this study.;In-vivo results indicate a statistically significant difference between the hemodynamic parameters calculated by the proposed deconvolution algorithm compared to those calculated by current deconvolution algorithms. By registering SWI images with perfusion images to exclude the large vessel structure, the hemodynamic parameters obtained were in excellent agreement with published values based on the microsphere technique, which is considered to be the gold standard. The techniques developed in this dissertation improve the accuracy of the calculation of hemodynamic parameters.