Objective: To systematically evaluate the accuracy of noncontrast computed tomography (NCT), perfusion computed tomography (PCT), and computed tomographic angiography (CTA) in determining site of occlusion, infarct core, salvageable brain tissue, and collateral flow in a large series of patients suspected of acute stroke.
Methods: We retrospectively identified all consecutive patients with signs and symptoms suggesting hemispheric stroke of < 48 hours in duration who were evaluated on admission by NCT, PCT, and CTA, and underwent a follow-up CT/CTA or magnetic resonance imaging (MRI) and magnetic resonance angiography (MRA) within 6 months of initial imaging. Two neuroradiologists evaluated NCT for hypodensity, PCT for infarct core and salvageable brain tissue, and CTA source images and maximal intensity projections for site of occlusion, infarct core, and collateral flow. Follow-up CTA and MRA were assessed for persistent arterial occlusion or recanalization. Follow-up CT and MRI were reviewed for final infarct location and volume, and used as a gold standard to calculate sensitivity (SE) and specificity (SP) of initial imaging.
Results: A total of 113 patients were considered for analysis, including 55 patients with a final diagnosis of stroke. CTA source images were the most accurate technique in the detection of the site of occlusion (SE = 95%; SP = 100%). Decreased cerebral blood volume on PCT was the most accurate predictor of final infarct volume (SE = 80%; SP = 97%), Increased mean transit time on PCT was predictive of the tissue at risk for infarction in patients with persistent arterial occlusion. CTA maximal intensity projections was the best technique to quantify the degree of collateral circulation.
Interpretation: The most accurate assessment of the site of occlusion, infarct core, salvageable brain tissue, and collateral circulation in patients suspected of acute stroke is afforded by a combination of PCT and CTA.