Diagnostic performance of Radiological Society of North America structured reporting language for chest computed tomography findings in patients with COVID-19

Purpose: To evaluate the diagnostic performance of the RSNA structured reporting language for chest CT findings in patients with COVID-19.

Material and methods: Patients with suspected COVID-19 who underwent chest CT and RT-PCR tests were enrolled consecutively in this retrospective study, regardless of symptoms. Imaging findings were categorized as "typical", "indeterminate", "atypical", or "negative" according to RSNA reporting language and compared to RT-PCR. "Single, round GGO" and "single, peripheral GGO," do not fit the reporting language, were also analyzed as "indeterminate" patterns.

Results: Of the 1186 patients included in the analysis, the diagnosis of COVID-19 was confirmed in 388 patients. Of the 388 patients, CT findings were categorized as "typical" in 248, "indeterminate" in 77, and "negative" in 63. The sensitivity, specificity, and accuracy of "typical" findings were 63.9, 99.0, and 87.5% for COVID-19, respectively. In addition to the "typical" findings, the highest diagnostic accuracy of 92.2% was achieved when the "single, peripheral GGO" and "single, round GGO" were considered to be CT-positive.

Conclusion: The RSNA reporting language has significant diagnostic performance for identifying COVID-19 pneumonia. CT findings that do not exactly fit the RSNA reporting language, such as "single, round GGO" and "single, peripheral GGO" improve diagnostic performance.