Automated perfusion-weighted MRI metrics via localized arterial input functions

Abstract

This thesis describes and validates a new method for calculating perfusion-weighted MRI (PWI) metrics, a non-invasive technique for calculating cerebral blood flow by tracking a bolus of contrast agent. Past methods to do this calculation require human intermediaries and can lead to errors in the presence of delay and dispersion of the contrast bolus, situations which occur commonly in the pathological conditions which require PWI. The new method described calculates perfusion metrics by defining an arterial input function (AIF) for every voxel in the brain based upon the voxels in close proximity to it. This allows for automated calculation of perfusion metrics, and the localized nature of the AIFs creates an implicit regard for delay and dispersion. This thesis demonstrates that this local AIF method is indeed able to correct flow misestimations due to delay and dispersion, and that it is also more useful for predicting tissue outcome post-stroke.