Article Text

Download PDFPDF
Abstracts from the Association of British Neurologists Annual Meeting 2011
006 Derivation and evaluation of thresholds for tissue at risk in stroke using CT perfusion
  1. F McVerry,
  2. N Macdougall,
  3. J M Wardlaw,
  4. M J Macleod,
  5. K W Muir
  1. Institute of Neuroscience and Psychology, University of Glasgow, UK
  2. Ninewells Hospital Dundee, UK
  3. Western General Hospital, University of Edinburgh, UK
  4. University of Aberdeen, UK


Introduction CT Perfusion (CTP) may provide information on brain viability in stroke, but thresholds which define tissue at risk are based on small patient samples.

Methods CTP from 19 acute stroke patients was evaluated. Follow-up imaging was co-registered to CTP permitting accurate assessment of outcome in the regions evaluated. 65 perfusion thresholds were applied to CTP scans and compared to final infarct volume using receiver operator curves. The best thresholds along with a commonly used threshold from the literature were compared to infarct volume in an independent group using Bland-Altman plots.

Results Absolute mean transit time (MTT) (Area under the Curve {AUC}=0.81), relative MTT (AUC=0.81) and delay time (AUC=0.78) were best predictors of tissue at risk of infarction. After testing in an independent group, Delay time of +2 s was the best predictor of infarct volume (95% agreement levels =−44 to +30 mls, Bias =−6.9).

Conclusion Time-based perfusion thresholds work well as predictors of infarct volume. Delay of +2 s was the best predictor when tested in an independent cohort and was superior to the widely used threshold of relative MTT of 145%. Evaluation in larger populations is needed to assess the performance of viability thresholds.

Statistics from


  • Email: ferghalmcverry{at}

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.