Optimal CT perfusion thresholds for core and penumbra in acute posterior circulation infarction

Leon Stephen Edwards, Cecilia Cappelen-Smith, Dennis Cordato, Andrew Bivard, Leonid Churilov, Longting Lin, Chushuang Chen, Carlos Garcia-Esperon, Kenneth Butcher, Tim Kleinig, Phillip M.C. Choi, Xin Cheng, Qiang Dong, Richard I. Aviv, Mark William Parsons, on behalf of the INSPIRE Study Group

Research output: Contribution to journalArticleResearchpeer-review

4 Citations (Scopus)


Background: At least 20% of strokes involve the posterior circulation (PC). Compared to the anterior circulation, posterior circulation infarction (POCI) are frequently misdiagnosed. CT perfusion (CTP) has advanced stroke care by improving diagnostic accuracy and expanding eligibility for acute therapies. Clinical decisions are predicated upon precise estimates of the ischaemic penumbra and infarct core. Current thresholds for defining core and penumbra are based upon studies of anterior circulation stroke. We aimed to define the optimal CTP thresholds for core and penumbra in POCI. Methods: Data were analyzed from 331-patients diagnosed with acute POCI enrolled in the International-stroke-perfusion-registry (INSPIRE). Thirty-nine patients with baseline multimodal-CT with occlusion of a large PC-artery and follow up diffusion weighted MRI at 24–48 h were included. Patients were divided into two-groups based on artery-recanalization on follow-up imaging. Patients with no or complete recanalisation were used for penumbral and infarct-core analysis, respectively. A Receiver operating curve (ROC) analysis was used for voxel-based analysis. Optimality was defined as the CTP parameter and threshold which maximized the area-under-the-curve. Linear regression was used for volume based analysis determining the CTP threshold which resulted in the smallest mean volume difference between the acute perfusion lesion and follow up MRI. Subanalysis of PC-regions was performed. Results: Mean transit time (MTT) and delay time (DT) were the best CTP parameters to characterize ischaemic penumbra (AUC = 0.73). Optimal thresholds for penumbra were a DT >1 s and MTT>145%. Delay time (DT) best estimated the infarct core (AUC = 0.74). The optimal core threshold was a DT >1.5 s. The voxel-based analyses indicated CTP was most accurate in the calcarine (Penumbra-AUC = 0.75, Core-AUC = 0.79) and cerebellar regions (Penumbra-AUC = 0.65, Core-AUC = 0.79). For the volume-based analyses, MTT >160% demonstrated best correlation and smallest mean-volume difference between the penumbral estimate and follow-up MRI (R2 = 0.71). MTT >170% resulted in the smallest mean-volume difference between the core estimate and follow-up MRI, but with poor correlation (R2 = 0.11). Conclusion: CTP has promising diagnostic utility in POCI. Accuracy of CTP varies by brain region. Optimal thresholds to define penumbra were DT >1 s and MTT >145%. The optimal threshold for core was a DT >1.5 s. However, CTP core volume estimates should be interpreted with caution.

Original languageEnglish
Article number1092505
Number of pages8
JournalFrontiers in Neurology
Publication statusPublished - 9 Feb 2023


  • cerebral perfusion
  • computerized tomography (CT)
  • CT perfusion (CTP)
  • ischaemic stroke
  • magnetic resonance imaging
  • posterior circulation infarction

Cite this