Introduction CT Perfusion (CTP) can improve diagnostic accuracy and support decision making in patients with acute ischemic stroke (AIS). The usability of CTP software packages is still limited by unreliability of results.

Aim of Study To compare core volumes estimate by Cercare Stroke (CP) and syngo.via (SV) with the core volumes segmented on non-contrast CT-scans.

Methods We selected patients with AIS with M1 occlusion having undergone thrombectomy with TICI 3 result in our clinic. CTP datasets were post-processed by CP and SV. Default settings were used for CP and three different settings for SV: default setting (A), additional smoothing (B) and adjusted settings (C). The results were compared to the core volume semi-automatically segmented in follow-up non-contrast CT. Agreement with the core volume was assessed using the intraclass coefficient (ICC) and the Mann-Whitney-U-test (MWU).

Results Sixty-six patients were included. Median segmented core volume was 6.4 ml (range 0–264 ml). Median core volume was 10.7 ml with CP, 41.1 ml with SV method A, 19.9 ml with SV B, and 29.6 ml with SV C. Agreement based on ICC was good for CP and SV method B and poor for SV methods A and C. MWU was 0.764 for CP and 0.0 for all SV results. The bias was smallest for CP (-6.8 ml).

Conclusion Core volume estimations and accuracy vary significantly between CTP software packages and should be acknowledged. Best agreement with segmented infarct volumes was provided by CP. SV method B with smoothing showed the best results for syngo.via.

Disclosure of Interest nothing to disclose