Comparison of PMAneo and RAPID in acute ischemic stroke: agreement in CT perfusion analysis and DEFUSE 3 imaging eligibility

Purpose Automated computed tomography perfusion (CTP) analysis plays a key role in triaging acute ischemic stroke (AIS). This study aimed to evaluate the agreement of PMAneo compared with RAPID in a multicenter setting. Materials and methods We retrospectively reviewed 99 patients with anterior circulation AIS from three stroke centers using four different CT scanners. CTP data were processed using both PMAneo and RAPID. Volumetric agreement for ischemic core volume and Tmax > 6 s volume was assessed using intraclass correlation coefficients (ICCs) and Bland–Altman analysis. Multivariable regression analysis was performed to identify factors associated with volume discrepancies. EVT imaging eligibility was evaluated according to the DEFUSE 3 criteria (ischemic core volume < 70 mL, mismatch ratio ≥ 1.8, and absolute mismatch volume ≥ 15 mL), using RAPID as the comparator. Results Ischemic core volumes showed excellent agreement between the two platforms (ICC = 0.881, 95% CI 0.827–0.918), and Tmax > 6 s volumes showed good agreement (ICC = 0.761, 95% CI 0.283–0.896), whereas mismatch volumes showed moderate agreement (ICC = 0.661, 95% CI 0.146–0.844). For the ischemic core threshold (< 70 mL), PMAneo demonstrated 100% (95% CI 96–100%) relative sensitivity and 67% (95% CI 38–88%) relative specificity. For the full DEFUSE 3 criteria, the relative sensitivity and specificity were 81% (95% CI 70–89%) and 84% (95% CI 66–95%), respectively. Bland–Altman analysis demonstrated proportional bias, with PMAneo underestimating volumes relative to RAPID as lesion size increased. Basal ganglia involvement and imaging site were significantly associated with these volume differences ( p < 0.05). Conclusion PMAneo showed high volumetric agreement and comparable DEFUSE 3 imaging eligibility to RAPID, suggesting it is a reliable alternative for automated CTP analysis. Although proportional bias and potential diagnostic discrepancies were observed in borderline cases near clinically relevant thresholds, PMAneo may facilitate broader access to automated CTP analysis.