Neck CT angiography in acute stroke: An open window for fast detection of COVID-19 lung involvement? Applicability in telemedicine

BackgroundChest CT has been proposed as a screening test to rule out SARS-CoV-2 lung infection in acute stroke. Our objectives are to analyze the predictive value of neck CT angiography (CTA) source images compared with conventional chest CT, the interobserver concordance and the reliability of the diagnosis using a mobile app. MethodsA retrospective observational study that included acute stroke patients admitted to a stroke center. Two raters blinded to the clinical data evaluated and classified the pulmonary findings in chest CT and neck CTA source images according to the COVID-19 Reporting and Data System (CO-RADS). CTA findings were evaluated using a conventional workstation and the JOIN mobile app. Scores of 3-5 were grouped as appearing typical or indeterminate for COVID-19 lung involvement and 0-2 as appearing atypical or negative for pneumonia. SARS-CoV-2 infection was confirmed by polymerase chain reaction (PCR). ResultsA total of 242 patients were included (42 with PCR-confirmed COVID-19). In the cohort of 43 patients with both neck CTA and chest CT, the predictive value for COVID-19 was equivalent (sensitivity, 53.8%; specificity, 92.9%). The interobserver agreement in the classification into CO-RADS 3-5 or 1-2 in CTA was good (K = 0.694; standard error, 0.107). In the cohort of 242 patients with neck CTA, the intraobserver agreement between the workstation and the JOIN app was perfect (K = 1.000; standard error 0.000). ConclusionsNeck CTA enables the accurate identification of COVID-19-associated lung abnormalities in acute stroke. CO-RADS evaluations through mobile applications have a predictive value similar to the usual platforms.

Automated summary

This study compared the predictive value of neck CTA source images and conventional chest CT to rule out SARS-CoV-2 lung infection. The results showed that neck CTA accurately identified COVID-19-associated lung abnormalities in acute stroke patients. Mobile applications for CO-RADS evaluations had a predictive value similar to traditional platforms.