Structured reporting of high-resolution CT based on ILD-RADS and its clinical application evaluation

YIN Jianbing; LIU Jia; MING Jialei; ZHANG Yanan; CUI Lei

doi:10.19300/j.2024.L21317

Abstract：Objective To assess the consistency of structured reports generated by different radiologists based on high-resolution CT Interstitial Lung Disease Reporting and Data System(ILD-RADS),and to analyze the assessment of these reports by rheumatologists.Methods Chest HRCT imaging and clinical data of 60 patients with clinically diagnosed interstitial lung disease(ILD)were retrospectively analyzed.A structured reporting template was developed according to relevant literature and guidelines.The HRCT features were categorized into fibrotic features,parenchymal abnormalities,and other findings.The presence of ILD and ILD-RADS grading were determined:grade 0(incomplete evaluation),ILD-RADS 1(typical usual interstitial pneumonia[UIP]),ILD-RADS 2(probable UIP),ILD-RADS 3(indeterminate UIP),and ILD-RADS 4(non-UIP).Two radiologists independently generated structured reports,while traditional narrative reports were directly exported from the reporting system.Three rheumatologists rated the completeness and clarity of both reporting styles using a Likert scale.Inter-observer consistency for the structured reports was analyzed using the Kappa test,and differences in ratings between the two reporting methods were compared using the rank-sum test.Results The structured reporting evaluations by the two radiologists demonstrated high consistency for ILD-RADS grading(kappa values for grades 1-4:0.812,0.839,0.856,and 0.861;overall kappa=0.878;all kappa>0.80)and moderate-to-high consistency for HRCT feature assessments(all kappa>0.60).The three rheumatologists uniformly rated the structured reports as 10/10 for both completeness and clarity.Scores for the structured reports were significantly higher than those for traditional narrative reports(all P<0.05).Conclusion Structured reporting based on ILD-RADS can provide clear,completely and consistent image descriptions of imaging findings with high inter-observer consistency.This approach demonstrates strong practicality in the evaluation of ILD.