CT perfusion imaging has become established as a method of choice for the assessment of ischemic stroke. However, current CT perfusion analysis methods have several shortcomings and do not exploit the imaging technique to its full potential. This thesis presents a model-based analysis method that is intrinsically robust and flexible. The method may furthermore aid in predicting hemorrhagic transformation; an adverse effect of thrombolytic stroke treatment. The application of this method to thin slice, high-resolution scans may well improve the diagnostic and prognostic value of CT perfusion imaging.
The proposed model-based method improves on the current commercial and academic state-of-the-art methods. By its simplicity and flexibility, it provides CBF, CBV, and MTT estimates superior to bSVD, and permeability estimates superior to Patlak analysis. Analysis by NLR using a simple, box-shaped IRF was found to be fast enough for use in a time-critical clinical setting. Even model-based quantitative analysis of thin slice CTP scans is feasible, improving the sensitivity in detection of small volume strokes, although this requires revision of currently used thresholds on absolute CBV, CBV, or MTT values for infarct and penumbra delineation. The advances in CTP analysis will in time provide more accurate and more in-depth clinical assessment possibilities for patients with acute ischemic stroke.