A Comparison of CT Perfusion Output of Rapid.AI and Viz.ai software in the Evaluation of Acute Ischemic Stroke

BACKGROUND AND PURPOSE: Automated CT Perfusion post-processing packages have been developed for managing acute ischemic stroke (AIS). These packages identify the volume of the ischemic core and penumbra by using advanced image processing techniques. This study aims to investigate the agreement of decision-making rules and output values derived from RapidAI and Viz.ai software packages in early and late time windows and to identify predictors of inadequate quality CT perfusion (CTP) studies.

MATERIALS AND METHODS: 129 AIS patients who had CTP performed upon presentation were analyzed. Imaging data were processed by two software packages: RapidAI and Viz.ai. Volumetric outputs were compared between packages by performing Spearman rank-order correlation and Wilcoxon signed-rank tests with sub-analysis performed at early (<6 hours) and extended (>6 hours) time windows. The concordance of selecting patients based on DAWN and DEFUSE3 eligibility criteria was assessed using Mcnemar test.

RESULTS: 108 out of 129 patients were found to have adequate quality studies. Spearman rank-order correlation coefficients were calculated on Tmax >6s volume, Tmax >10s volume, CBF <30% volume, Mismatch Volume, and Mismatch Ratio, between both software packages with correlation coefficients of 0.82, 0.65, 0.77, 0.78, 0.59 respectively. The Wilcoxon Signed-Rank Test was also performed on Tmax >6s volume, Tmax >10s volume, CBF <30% volume, Mismatch Volume, and Mismatch Ratio with P-Values of 0.30, 0.016, <0.001, 0.03, <0.001 respectively. In a one-sided test, CBF <30% was greater in Viz.ai (p<0.001). Although this resulted in statistically significant differences, it did not cause clinically significant differences when applied to the DAWN and DEFUSE 3 criteria. Lower ejection fraction (EF) predicted an inadequate study in both software packages (P = 0.018; 95% CI: 0.01, 0.113) and (P = 0.024; 95% CI: 0.008, 0.109); for RapidAI and Viz.ai, respectively. In Viz.ai, the presence of a clip, coil, or other metal predicted an inadequate study (P = 0.042; 95% CI: -3.225, -0.057).

CONCLUSIONS: Viz.ai predicted higher ischemic core volumes than RapidAI. Viz.ai predicted lower combined core and penumbra values than RapidAI at lower volumes and higher estimates than RapidAI at higher volumes. Clinicians should be cautious when using different software packages for clinical decision-making.

ABBREVIATIONS: LVO =Large vessel occlusion; time to maximum peak = Tmax; AIS = acute ischemic stroke; LKW = last known well;