Lacunar syndromes correlate with a lacunar stroke on imaging in 50% to 60% of cases. Computed tomography perfusion (CTP) is becoming the preferred imaging modality for acute stroke triage. We aimed to estimate the sensitivity, specificity, and predictive values for noncontrast computed tomography and CTP in lacunar syndromes, and for cortical, subcortical, and posterior fossa regions. A retrospective analysis of confirmed ischemic stroke patients who underwent acute CTP and follow-up magnetic resonance imaging between 2010 and 2018 was performed. Brain noncontrast computed tomography and CTP were assessed independently by 2 stroke neurologists. Receiver operating characteristic curve analysis was performed to estimate sensitivity, specificity, and area under the curve (AUC) for the detection of strokes in patients with lacunar syndromes using different CTP maps. We found 106 clinical lacunar syndromes, but on diffusion-weighted imaging, these consisted of 59 lacunar, 33 cortical, and 14 posterior fossa strokes. The discrimination of ischemia identification was very poor using noncontrast computed tomography in all 3 regions, but good for cortical (AUC, 0.82) and poor for subcortical and posterior regions (AUCs, 0.55 and 0.66) using automated core-penumbra maps. The addition of delay time and mean transient time maps substantially increased subcortical (AUC, 0.80) and slightly posterior stroke detection (AUC, 0.69). Analysis of mean transient time and delay time maps in combination with core-penumbra maps improves detection of subcortical and posterior strokes.
Read full abstract