Harmonizing and Optimizing CT Perfusion Stroke Imaging

Abstract

This thesis focuses on harmonizing and optimizing CT perfusion (CTP) imaging for stroke. CTP imaging can help select patients with ischemic stroke for thrombectomy. However, due to a lack of consensus on the CTP acquisition and processing protocols, widespread acceptance of image-based treatment criteria has not been achieved. The CLEOPATRA care assessment was set up to determine whether it is cost-effective to select patients with ischemic stroke for thrombectomy based on stroke imaging. We focused on harmonizing and optimizing the protocols for CTP stroke imaging. Chapter 1 introduces some concepts of CTP imaging. In CTP imaging, contrast agent is injected into the bloodstream. By measuring changes in contrast enhancement over time, perfusion maps can be created. These perfusion maps show the blood flow in brain tissue and are used to estimate ischemic (less perfused) brain regions. The estimated ischemic regions help clinical decision-making for ischemic stroke. Chapters 2 and 3 discuss the injection protocol, scan protocol, and processing protocol of CTP imaging. Using an anthropomorphic digital CTP phantom, we argued that the processing protocol has the greatest impact on CTP imaging results. Additionally, variations in the injection and scan protocols between centers can, in some cases, lead to different CTP imaging results. Chapter 4 describes our efforts to create a physical CTP phantom from the digital phantom in chapters 2 and 3. By scanning paper sheets printed with contrast agent, we demonstrated that simulating anthropomorphic brain tissue perfusion with a physical phantom is feasible. Chapters 5 and 6 cover the estimation of ischemic regions. A standardized method that we developed could harmonize and optimize CTP imaging results to some extent. Additionally, the estimations from this standardized method resembled both manual segmentations of CTP perfusion maps and acute MR diffusion images. Chapter 7 presents a new use of CTP imaging as a tool to locate vessel occlusions. Smaller and more distal occlusions are increasingly difficult to locate with CT angiography imaging. We demonstrated that the location of the total ischemic region, determined from CTP imaging, can help locate the vessel occlusion. Chapter 8 recommends establishing a standardized framework for estimating ischemic regions to reduce inconsistencies in clinically relevant information derived from CTP imaging. Based on our research, this framework should utilize all perfusion data, consider spatial information, and rely on manual segmentations as a reference point. In summary, we believe that CTP imaging remains important for helping patients with ischemic stroke. This thesis argues that current inconsistencies arise from misinterpretations of perfusion maps. We advocate for a shifted perspective in which CTP imaging is assessed on its intrinsic value and broader clinical applications are considered. Only then can the full potential of CTP imaging be realized.