Abstract 2450: Assessing The Variability Of Ctp Post Processing Techniques To Define The Acute Infarct Core And Penumbra In A Large Dataset
Aim: In order to fully validate CTP for the selection of acute stroke patients for treatment, comparison of the accuracy of different deconvolution techniques to detect infarct core and penumbra is required.
Methods: A cohort of 314 patients that presented with <6 hour hemispheric ischemia between 2004 and 2011 were studied retrospectively. Patients were imaged with multimodal CT at baseline and MRI at 24 hours. From the patient cohort, 133 had no significant reperfusion at 24 hours, so were suitable for the analyses to define the ischemic penumbra. Another 102 patients with major reperfusion at 24 hours, were suitable for the analyses to define the acute infarct core. Perfusion CT maps were generated using the Maximum slope (Peters) model, Partial Deconvolution (PD), Single Value Deconvolution (SVD), Single Value Deconvolution with Delay Correction (cSVD) and a Block Circulant Deconvolution (BCD). The perfusion CT maps were coregistered to the corresponding acute and 24 hour DWI. Receiver Operating Characteristic (ROC) Curve Analysis was used to test the predictive performance of CTP in relation to the DWI infarct core.
Results: Using the Maximum slope model, a relative CBF 145% of normal (AUC 0.71) was demonstrated the best threshold to define the acute penumbra. Using the partial deconvolution approach CBF 155% of normal (AUC 0.73) and a TTP <4 seconds (AUC 0.74). The threshold to describe the acute infarct core with SVD was a CBF 6 seconds (AUC 0.77). Using SVD with delay correction a CBF 2 seconds +normal (AUC 0.81) accurately defined the acute penumbra. Using a block circulant method the infarct core was best defined by CBF <15 mL/100g/min (AUC 0.69) and a DT < 4 seconds (+baseline) is the most accurate when defining the acute penumbra (AUC 0.72).
Discussion: cSVD was the most accurate method of defining both the acute infarct core and penumbra in this study. The SVD, BCD and max slope methods all produced different thresholds to define the acute stroke pathophysiology. Of note is that a CBF was always the best method to define the acute infarct core, regardless of the method. However the threshold to define the acute infarct core did vary. This is the first study to ever directly compare the acute threshold for tissue pathophysiology in ischaemic stroke with different CTP post processing methods.
- © 2012 by American Heart Association, Inc.