Comparison of Automated CT Perfusion Softwares in Evaluation of Acute Ischemic Stroke

Background and Purpose: Automated imaging software is integral to decision-making in acute ischemic stroke (AIS) during extended time windows. RAPID software is the most widely used and has been validated in landmark endovascular trials. Olea software is another commercially available and FDA-approved software, but has not been studied in AIS trials. We aimed to compare the diagnostic utility and accuracy of RAPID and Olea in everyday clinical practice outside of a clinical trial. Methods: We analyzed prospectively-collected data from a consecutive cohort of 141 patients with suspected AIS who underwent computed tomography perfusion upon presentation followed by diffusion-weighted magnetic resonance imaging (DWI-MRI) within 24-48 hours. Core infarct was defined as the region with a relative cerebral blood flow (rCBF) less than 30% on RAPID and rCBF less than 40% on Olea (default settings). We also evaluated rCBF less than 30% on Olea to match RAPID's default setting. Infarct volume on DWI-MRI was measured using a semi-automated segmentation method. Results: Twenty-one patients were excluded; 14 due to poor bolus tracking and/or motion artifact, and 7 due to software failure. The software failure rate was 4.7% [6/127] with RAPID versus .78% [1/127] with Olea (P = .12). For the remaining 120 patients, the sensitivity and specificity for detecting an acute infarct were 40.5% and 97.6% for RAPID; 50.6% and 85.4% for Olea; and for detecting large infarcts (>= 70 mL on DWI-MRI) 73.7% and 81.2% for RAPID; 73.7% and 68.3% for Olea. Core infarct volume on RAPID was more closely correlated with DWI-MRI infarct volume (rho = .64) than Olea (rho = .42). Conclusions: Our head-to-head comparison of these 2 commonly-used softwares in the clinical setting elucidates the pros and cons of their use to guide decision-making for AIS management in the acute setting.