Development and evaluation of CT acquisition and analysis methods of brain perfusion

CTP has shown to be a promising tool for selection of patients with ischemic stroke who can potentially benefit from administration of rtPA. Although CTP is fast, more widely available, and cheaper compared to other imaging modalities such as MRI and PET scans, it has not yet gained widespread acceptance in this role. The different tissue perfusion parameter maps may indicate widely different lesion size and locations, this is the result of the larger influence of technical and biological factors on the estimation of tissue perfusion compared to the anatomical image that is provided by angiography. There is need for establishment of a viable protocol that is proven to yield accurate results and can be used consistently by clinicians and software vendors. In 2010 and 2013 Kudo et. al. demonstrated how this lack of consensus is causing an unacceptable variability in the perfusion measurements performed by different software. Kudo’s publication compared results from different software when given the same input, even stronger variation can be found when the one particular software is used to analyse the same perfusion but with different protocols. Attempts have been made by the CTP community to reach a consensus, but without standardization of the tools being used to assess the best protocol and analysis method it will be difficult to reach this consensus. Comparison with perfusion measurements using microsphere based measurements in animal studies and other modalities such as PET and MRI is possible, but not in the same patient. Almost simultaneous acquisition would be required to ensure the same conditions are measured for quantitative definition of accuracy of CTP measurements. Furthermore, since clinical CTP acquisitions cannot be repeated with equal or varying settings without administering excessive CT dose, effects of different noise or tube settings (mAs) cannot be determined. This hampers the dose and image quality optimization. This thesis describes methods to evaluate the impact of technical parameters from acquisition and analysis protocol and (non-clinical) patient specifics on CTP outcome in acute stroke. To this end a new hybrid brain phantom is suggested. Subsequently, specific attention is paid to the impact of de-convolution algorithms, patient motion and AIF location selection. This could help standardizing and optimizing CTP studies and increasing its clinical impact.