Atrial fibrillation detected after stroke (AFDAS) may be either newly emergent or previously undiagnosed, with potential neurogenic origin. The ability to predict AFDAS through early imaging remains limited. This study investigates whether hyperacute CT perfusion imaging can identify patients at higher risk for AFDAS. We retrospectively analyzed data of patients with acute ischemic stroke admitted to the University Hospital of Trieste between January 2019 and March 2020. Patients with known AF, hemorrhagic stroke, or TIA were excluded. Lesion volumes and locations were assessed using perfusion and follow-up CT imaging. Among the 242 patients included, AFDAS was more frequently identified in those of older age, female sex, with higher NT-proBNP levels, and worse functional outcomes. Frontal-parietal-insular hypoperfusion was more prevalent in patients with AFDAS. After multiple adjustments, frontal lobe hypoperfusion remained independently associated with AFDAS (OR 2.836, 95% CI 1.406–5.719, p=0.004), while no significant associations were found with final infarct volume or location on follow-up imaging. This finding was confirmed in a sub-analysis restricted to non-lacunar strokes (OR 2.286, 95% CI 1.068–4.894, p=0.033). NT-proBNP levels, non-current smoking status, and frontal hypoperfusion on hyperacute CT perfusion imaging may predict AFDAS, suggesting that early central autonomic dysregulation, rather than infarcted tissue, plays a role in the pathogenesis of atrial fibrillation. These findings support the integration of perfusion imaging and cardiac biomarkers into AFDAS risk stratification models; however, validation through prospective multicenter studies is warranted.
