This study aimed to identify and validate whole brain perfusion computed tomography (CTP) thresholds for ischemic core and salvageable penumbra in acute stroke patients and develop a probability based model to increase the accuracy of tissue pathophysiology measurements.

One hundred and eighty-three patients underwent multimodal stroke CT using a 320-slice scanner within 6hours of acute stroke onset, followed by 24hour MRI that included diffusion weighted imaging (DWI) and dynamic susceptibility weighted perfusion imaging (PWI). Coregistered acute CTP and 24hour DWI was used to identify the optimum single perfusion parameter thresholds to define penumbra (in patients without reperfusion), and ischemic core (in patients with reperfusion), using a pixel based receiver operator curve analysis. Then, these results were used to develop a sigma curve fitted probability based model incorporating multiple perfusion parameter thresholds.

For single perfusion thresholds, a time to peak (TTP) of +5seconds best defined the penumbra (area under the curve, AUC 0.79 CI 0.74–0.83) while a cerebral blood flow (CBF) of < 50% best defined the acute ischemic core (AUC 0.73, CI 0.69–0.77). The probability model was more accurate at detecting the ischemic core (AUC 0.80 SD 0.75–0.83) and penumbra (0.85 SD 0.83–0.87) and was significantly closer in volume to the corresponding reference DWI (P=0.031).

Whole brain CTP can accurately identify penumbra and ischemic core using similar thresholds to previously validated 16 or 64 slice CTP. Additionally, a novel probability based model was closer to defining the ischemic core and penumbra than single thresholds.