Performance and reliability comparison: original vs. revised bone reporting and data system (Bone-RADS).

Abstract

Objective: To propose a revised bone reporting and data system (Bone-RADS) and evaluate its diagnostic performance and inter-reader reliability compared to the original Bone-RADS for solitary bone lesions on CT.

Materials and methods: This retrospective study included 159 adult patients (mean age: 56 ± 19 years; 88 men) who underwent bone biopsy for solitary bone lesions between March 2005 and September 2021. Two radiologists (R1/2) independently categorized the lesions twice, once using the original Bone-RADS and once using the revised version. Lesions were classified as follows: (1, benign; 2, incompletely assessed; 3, indeterminate; 4, malignancy or requiring treatment). The revised Bone-RADS excluded the original criteria for lesion related pain and history of malignancy. Diagnostic performance was assessed using histopathology as the reference standard, and inter-reader reliability was analyzed.

Results: The bone lesions included 96 lucent and 63 sclerotic/mixed lesions. Sensitivity showed no significant difference between the original and revised Bone-RADS for both readers across lucent and sclerotic/mixed lesions (all P ≥ .05). However, the specificity of the revised Bone-RADS was significantly higher than that of the original (lucent: 11% vs. 50% [R1], 11% vs. 46% [R2]; sclerotic/mixed: 32% vs. 92% [R1], 32% vs. 86% [R2]). Other performance metrics, including positive/negative predictive value and accuracy, were also higher in the revised Bone-RADS. Inter-reader reliability was higher for the revised Bone-RADS compared to the original (κ = .744 vs .854).

Conclusion: The revised Bone-RADS significantly improved specificity while maintaining sensitivity compared to the original version.