MRI Perfusion Measurements using Magnetic Susceptibility Effects: : Calibration Approaches and Contrast Agent Quantification

Sammanfattning: Exchange of oxygen and nutrients between blood and tissue occurs at the capillary level of the blood system. The blood volume flow rate in the capillaries is often referred to as perfusion, and knowledge about perfusion provides important information about the function and viability of the tissue, for example, in patients with ischaemic stroke, cancer, and neurodegenerative diseases.Dynamic susceptibility contrast magnetic resonance imaging (DSC–MRI) is the basic data collection approach used to obtain perfusion information from the studies reported in this doctoral thesis. The approach shows the advantage of providing estimates of not only brain perfusion, or cerebral blood flow (CBF), but also cerebral blood volume (CBV) and mean transit time (MTT). All of these parameters are required for a comprehensive description of microcirculation. When using standard implementations of DSC–MRI image acquisition and post-processing routines, the CBF and CBV values are often overestimated. In DSC–MRI, a contrast agent (CA) bolus is injected intravenously in an arm vein and is subsequently tracked by rapid imaging when passing through the brain. To calculate CBF, CBV, and MTT, CA concentration information from both tissue and blood is required. The main problem of DSC–MRI is estimating reliable CA concentration levels in tissue and artery simultaneously. Relatedly, if transverse relaxivity-based information is used, then the response to the CA, in terms of the change in relaxation rate, ΔR2∗, differs between blood and tissue. Additionally, there is a non-linear dependence on CA concentration in whole blood. Another issue related to estimations of blood concentration is the practical difficulty of finding a voxel containing only blood, implying that the concentration time curve representing blood will be affected by the surrounding tissue, possibly influencing both the shape of the curve and the absolute levels of estimated concentrations. One approach is to obtain CBV estimates with alternative methods, where contrast agent concentration is quantified using the longitudinal relaxation time, T1, or by using a special MRI pulse sequence designed to study the blood contribution to the MRI signal. The information from these additional CBV estimates is used to calibrate the conventional DSC–MRI data. This approach provided perfusion results with the expected absolute levels but showed moderate repeatability and low correlation with arterial spin labelling (ASL), used as a reference perfusion imaging method. Another study included in the context of this doctoral thesis dealt with the issue of partial-volume effects in the voxel selected to represent blood. By rescaling the area under the curve (AUC) of the concentration-versus-time curve measured in an artery (assumed to show the desired shape), with the AUC of a large vein, measured under circumstances more favourable for blood-data registration, CBF and CBV values were calibrated. This method aims primarily to correct for partial volume effects in the blood voxel (i.e. it does not directly address relaxivity issues). An observation of interest in this context was that absolute perfusion levels similar to what was expected from literature data were obtained. Furthermore, the repeatability was more promising using this approach, compared to the ones described above. The phase of the MRI signal is related to the magnetic field strength, which, in turn, is related to the magnetic susceptibility. As the CA alters the magnetic susceptibility, it should, in principle, be possible to obtain information about CA concentration using MRI phase information. This approach was used in three of the studies described in this thesis, for calculation of perfusionrelated parameters in relative and absolute terms. These studies indicated that information about (or related to) magnetic susceptibility is a promising method for estimating CA concentration, whereas a number of methodological issues still need to be resolved or further investigated. It was also shown that the shape of an arterial ΔR2' curve, based on DSC–MRI data, displayed a shape similar to the corresponding curve obtained by using magnetic susceptibility information for assessment of concentration. Thus, the shape of a typical arterial blood concentration curve used in a standard DSC–MRI experiment can probably be regarded reasonably reliable. However, the AUC is likely to be underestimated because of partial-volume effects.