Inter-observer and intra-observer agreement of bone reporting and data system (Bone-RADS) in the interpretation of bone tumors on computed tomography

Abstract

Objective

To assess inter-observer & intra-observer agreement of the American College of Radiology (ACR) bone reporting and data system (Bone-RADS) in the interpretation of bone tumors on computed tomography (CT).

Methods

This retrospective study included 273 bone tumors 184 (67.4 %) benign and 89 (32.6 %) malignant. Two blinded radiologists independently reviewed the CT images to assess the defined CT features of bone lesions and assign a bone-RADS category. A third observer reviewed the CT images twice with one month interval and reported the specified CT features and final bone-RADS category for bone lesions. Inter-observer and intra-observer agreement of bone-RADS were analyzed.

Results

There was almost perfect inter-observer agreement between the two reviewers for all defined variables as well as bone-RADS categories (Kappa = 92.2 %). Overall intra-observer agreement was also perfect for all defined CT features and bone RADS categories with higher percentage than inter-observer one (κ = 98.3 %).

Conclusion

Bone-RADS is an effective clinical tool for practicing radiologists in the risk assessment and management of bone tumors with perfect agreement among observers. To our knowledge, no studies have been conducted for assessing reliability or validity of the recent ACR Bone-RADS using CT. Hence, this study could serve as a cornerstone for further upcoming studies to assess the reproducibility and validity of the ACR Bone-RADS.