Interobserver agreement of interstitial lung fibrosis Reporting and Data System (ILF-RADS) at high-resolution CT.

PURPOSE

To assess the interobserver agreement of interstitial lung fibrosis Reporting and Data System (ILF-RADS) in interpretation and categorization of interstitial lung disease (ILD) at high-resolution CT (HRCT).

METHODS

Retrospective analysis was performed on 65 consecutive patients (36 male and 29 female), median age 53 years, who were referred to the Radiology Department, Mansoura University, in the period from July 2016 to February 2020. They were expected clinically to have diffuse lung disease and underwent HRCT of the chest. Patients had some investigations like serology, and when required surgical lung biopsy. Image analysis was done by two independent and blinded readers for the pulmonary and extra-pulmonary finding of ILF-RADS. The pulmonary findings were 13 items and extrapulmonary findings were 5 items. The score was 5 types according to ILF-RADS: ILF-RADS 0 (incomplete assessment), ILF-RADS 1 (typical UIP), ILF-RADS 2 (probable UIP), ILF-RADS 3 (indeterminate UIP), ILF-RADS 4 (CT features most consistent with non-UIP diagnosis).

RESULTS

There was an excellent interobserver agreement of both reviewers for overall ILF-RADS (K = 0.88, P = 0.001) with 95.4% agreement. There was an excellent interobserver agreement for overall pulmonary findings (K = 0.901, 95% CI = 0.877-0.926, P = 0.001), excellent interobserver agreement for seven items including lung volume, traction bronchiectasis, nodules, cysts, consolidation, emphysema, and complications and moderate interobserver agreement for six items including reticulations, honeycomb, ground glass, mosaic attenuation, and axial and zonal distribution. There was excellent interobserver agreement for overall extra-pulmonary findings (K = 0.902, 95% CI = 0.852-0.952, P = 0.001), excellent interobserver agreement for four items including mediastinum, pleura, visible abdomen, and soft tissue and bone and moderate interobserver agreement for trachea and main bronchi. There was excellent interobserver agreement for ILF-RADS score: ILF-RADS 1 (K = 0.84, P = 0.001), ILF-RADS 3 (K = 0.881, P = 0.001), and ILF-RADS 4 (K = 0.878, 95% CI = 0.743-1.0, P = 0.001) and moderate interobserver agreement for ILF-RADS 2 (K = 0.784, P = 0.001).

CONCLUSION

ILF-RADS is a reliable reporting system which can be routinely performed for standard interpretation of ILD.