Abstract TP69: Three-Phase Helical Computed Tomographic Angiography Perfusion Maps Predict Final Infarction in the Acute Ischemic Stroke Setting

Abstract

Background: Three-phase helical CTA can provide information on parenchymal hemodynamics distal to the occlusion, similar to CT perfusion (CTP). Helical CTA is a less expensive and a more widely available modality. We propose two perfusion map algorithms applied to three-phase helical CTA, providing optimal thresholds for prediction of final infarct volume Methods: 44 stroke patients with occlusion visible on CTA were acutely imaged with three-phase CTA (temporal sampling was 8 seconds). MR diffusion weighted imaging (DWI) between 24-48 hours were used to measure final infarct volume. The three-phase helical CTA perfusion maps were denoted “Delay” and “Flow Average” (see Figure 1). The maps were filtered using a 3D Gaussian blurring technique. The maps were generated for all patients, then a receiver operating characteristic (ROC) curve was generated for the merged patient data, comparing infarct vs. normal tissue. Thresholds were determined using ROC curves by optimizing sensitivity and specificity. Results: The “Delay” map generated an ROC curve with an Area-Under-Curve (AUC) of statistical significance. The “Flow Average” map generated an ROC curve with an Area-Under-Curve (AUC) of statistical significance. Conclusion: The proposed “Delay” and “Flow Average” perfusion maps applied to three-phase CTA predicted final infarct volume to a high degree of accuracy, close to CTP accuracies from the literature. These results show the capability of three-phase helical CTA to generate quantitative perfusion maps which will be useful for non-tertiary centers that do not have access to expensive post-processing software.