Inter-observer and inter-examination variability of manual vertebral bone attenuation measurements on computed tomography

Abstract

Objective: To determine inter-observer and inter-examination variability of manual attenuation measurements of the vertebrae in low-dose unenhanced chest computed tomography (CT). Methods: Three hundred and sixty-seven lung cancer screening trial participants who underwent baseline and repeat unenhanced low-dose CT after 3 months because of an indeterminate lung nodule were included. The CT attenuation value of the first lumbar vertebrae (L1) was measured in all CTs by one observer to obtain inter-examination reliability. Six observers performed measurements in 100 randomly selected CTs to determine agreement with limits of agreement and Bland-Altman plots and reliability with intraclass correlation coefficients (ICCs). Reclassification analyses were performed using a threshold of 110 HU to define osteoporosis. Results: Inter-examination reliability was excellent with an ICC of 0.92 (p <0.001). Inter-examination limits of agreement ranged from -26 to 28 HU with a mean difference of 1 ± 14 HU. Inter-observer reliability ICCs ranged from 0.70 to 0.91. Inter-examination variability led to 11.2 % reclassification of participants and inter-observer variability led to 22.1 % reclassification. Conclusions: Vertebral attenuation values can be manually quantified with good to excellent inter-examination and inter-observer reliability on unenhanced low-dose chest CT. This information is valuable for early detection of osteoporosis on low-dose chest CT. Key Points: • Vertebral attenuation values can be manually quantified on low-dose unenhanced CT reliably.• Vertebral attenuation measurements may be helpful in detecting subclinical low bone density.• This could become of importance in the detection of osteoporosis.