Abstract
<title>Abstract</title> <p>Computed tomography perfusion (CTP) imaging in acute ischemic stroke enables rapid and precise assessment of tissue perfusion status, supporting timely treatment decisions. We aimed to standardize optimal thresholds for infarct core, tissue at risk, and penumbra volumes across different CTP software packages using a digital perfusion phantom and to explore their relationship with stroke onset time to assess potential time-of-day variation. Digital phantom post-processing outputs from three Siemens Syngo.via packages and one GE Healthcare CT Perfusion 4D (CTP4D) software package were compared with those from iSchemiaView RAPID software. Equivalent relative cerebral blood flow (rCBF) and Time-to-Max (Tmax) thresholds corresponding to RAPID definitions were derived and retrospectively applied to multicenter clinical datasets. Phantom testing demonstrated inter-platform variability, with rCBF thresholds equivalent to RAPID ≤ 30% ranging from 15.0% to 20.0%, and Tmax thresholds equivalent to > 6 seconds ranging from 5.5 to 6.2 seconds. Among 208 patients across four sites, calibration reduced median core volume estimates and modestly increased penumbra (mismatch) volumes. The median core reduction was 31.8 mL (95% CI: −35.8 to − 23.8), while penumbra increased by 14.0 mL (95% CI: 8.9–18.3), although this effect varied across platforms. Consequently, the proportion of patients’ eligibles for mechanical thrombectomy increased by 39.2 percentage points (95% CI: 32.2 to 46.2). A diurnal trend was observed, with smaller core volumes in the early morning. These findings highlight the need for cross-platform CTP calibration to enable reliable multicenter comparisons.</p>