Diffusion and perfusion magnetic resonance imaging in human ischaemic stroke analysis strategies and measurement isues in the assessment of lesion evolution

Stroke is the third most common cause of death in the developed world, and is the major cause of long term disability in the U.K. Eighty percent of strokes are ischaemic, or caused by vascular occlusion. Magnetic resonance imaging (MRI) provides stroke researchers with an invaluable tool to visualise ischaemic processes non-invasively from very early times after stroke onset. However, despite major progress, it is still unclear to what degree particular appearances on MRI relate to the underlying pathophysiological state of the ischaemic lesion, or the clinical outcome of the stroke patient.The aim of this thesis is to explore the evolution and issues affecting the analysis of ischaemic lesions on diffusion- and perfusion-weighted MRI (DWI and PWI) from acute (< 24 hours) to chronic (90 days) times after stroke onset. This thesis includes a review of previous human studies of acute DWI and PWI appearance versus final lesion outcome, a review of the 'DWI/PWI mismatch' model (thought to represent the ischaemic penumbra, or 'tissue at risk' of infarction), and a systematic review of previous animal studies of the pathophysiology associated with particular lesion appearances on DWIThe methodological problems raised by these reviews are addressed in this thesis using a large cohort of stroke patients with serial DWI and PWI. The interrater variability of manual lesion measurements on acute DWI is investigated, and factors affecting this variability are discussed. The effect of lesion oedema (swelling) on measurements of ischaemic lesions on MRI is investigated. This thesis also investigates the tissue state underlying persistent hyperintensity on late DWI, and whether this is just T₂ 'shine through', or indicates distinct features in the evolution v of the lesion. A novel grid-based analysis method is developed and employed to track serial DWI and PWI changes more effectively, and the effect of observer variability on diffusion and perfusion parameters measured by this method is assessed. Lastly, this thesis discusses the concept of using 'threshold' values to predict tissue infarction or survival