Impact of calibration approach on image quality and volumetric bone mineral density in low-dose chest computed...

Opportunistic assessment of volumetric bone mineral density (vBMD) from computed tomography (CT) scans offers a strategy to identify individuals at risk of osteoporosis. However, accurate CT-based vBMD quantification depends on appropriate density calibration, and direct comparisons of synchronous, asynchronous, and tissue-based internal calibration in low-dose chest imaging are limited. This study had two objectives: (1) to assess the impact of calibration methods on image quality in low-dose chest CT, particularly whether the inclusion of a calibration phantom compromises image quality, and (2) to examine whether these variations influence agreement in vBMD measurements across calibration techniques. 51 participants underwent chest CT; 23 with a calibration phantom (synchronous) and 28 without (asynchronous). Trabecular vBMD was quantified at the fourth and eighth thoracic vertebrae (T4 and T8, respectively) and the first lumbar vertebra (L1). Image quality was assessed using stochastic noise and signal-to-noise ratio (SNR) in calibration phantom rods and tissue-based reference regions. Agreement in vBMD measurements was evaluated using Pearson's correlation coefficients, Bland-Altman analyses, and absolute deviations relative to internal calibration. Internal calibration tissues demonstrated stable image quality across synchronous and asynchronous acquisitions, whereas synchronously scanned calibration phantoms exhibited increased noise and reduced SNR. In vBMD measurements, asynchronous calibration demonstrated smaller deviations (19.15 ± 11.59 mg cm^-3vs. 10.19 ± 9.69 mg cm^-3) than synchronous across vertebral levels. These findings show that calibration strategy affects both image quality and vBMD agreement in low-dose chest CT. Asynchronous and tissue-based internal methods provided more consistent estimates than synchronous calibration, supporting their use for opportunistic osteoporosis assessment.