Cerebral perfusion software: comparative evaluation

Abstract

IntroductionPerfusion CT (CTP) is crucial in defining the best treatment strategy for patients with acute ischemic stroke. Several software is available for assessing cerebral perfusion maps, which may show some variability in results. We aim to investigate the agreement between CT perfusion values in acute ischemic stroke of 2 different software (Syngo.via and Icobrain) in estimating ischemic core, ischemic penumbra, and mismatch ratio.Methods17 patients (10 women, mean age = 79.5 years) with acute stroke with large vessel occlusion (LVO) were enrolled in our study between August 2022 and March 2023. All patients underwent a CTP scan to determine the best treatment approach: endovascular treatment or systemic fibrinolysis according to DAWN/DEFUSE-3 criteria. All CTP examinations were analyzed using 2 different dedicated software (Syngo.via, Icobrain) and the perfusion values obtained (core, penumbra, mismatch ratio) were compared with each other. A basal CT scan 24 h after treatment was acquired to assess the final infarct volume (FIV) and possible hemorrhagic infarction. Then we compared the true FIV, manually assessed, with the predicted FIV calculated by the two software.ResultsThe values obtained showed that the mean core volumes calculated by Syngo.via and Icobrain did not exhibit any statistically significant differences with a p-value of 0.09. There is no statistically significant difference between the mean penumbra volumes calculated by the two software (p-value = 0.29). There is no statistically significant difference between the mean values of core and penumbra calculated by both software. Furthermore, both Icobrain and Syngo.via gave the same therapeutic indication. Correlation analysis indicated a slightly stronger correlation between the core volume calculated by Icobrain and the FIV (r = 0.98 vs r = 0.94). Both correlations were statistically significant (p = < 0.001).ConclusionThe evaluation of perfusion indices by automated systems may present a certain degree of variability between different software, although they agree on the final therapeutic stratification.