Supplementary Material for: Does Perfusion Computed Tomography Facilitate Clinical Decision Making for Thrombolysis in Unselected Acute Patients with Suspected Ischaemic Stroke?

Background: Despite use in clinical practice and major positive trials of thrombolysis, non-contrast computed tomography (NCCT) is not sensitive for identifying penumbral tissue in acute stroke. This study evaluated how physiological imaging using CT perfusion (CTP) could add to the diagnostic utility of an NCCT and inform clinical decisions regarding thrombolysis. Methods: Forty imaging datasets containing NCCT and CTP were retrospectively identified from a cohort of consecutive acute stroke patients. Two sets of observers (n = 6) and a neuroradiologist evaluated the images without knowledge of clinical symptoms. Inter-observer agreement was calculated using the ĸ statistic for identifying acute ischaemic change on NCCT: perfusion abnormalities (namely cerebral blood volume, cerebral blood flow and time to peak), and penumbral tissue on perfusion maps obtained by two image processing algorithms. Results: Inter-rater agreement was moderate (ĸ = 0.54) for early ischaemic change on NCCT. Perfusion maps improved this to substantial for cerebral blood volume (ĸ = 0.67) and to almost perfect for time to peak (ĸ = 0.87) and cerebral blood flow (ĸ = 0.87). The agreement for qualitative assessment of penumbral tissue was substantial to perfect for images obtained using the two different perfusion algorithms. Overall, there was a high rate of decision to thrombolyse based on NCCT (81.25%). CTP strengthened the decision to thrombolyse based on NCCT in 38.3% of cases. It negatively influenced the decision in 14.6% of cases, this being significantly more common in experienced observers (p = 0.02). Conclusions: We demonstrate that the qualitative evaluation of CTP produces near perfect inter-observer agreement, regardless of the post-processing method used. CTP is a reliable, accessible and practical imaging modality that improves confidence in reaching the appropriate diagnosis. It is particularly useful for less experienced clinicians, to arrive at a physiologically informed treatment decision.