Visual grading of motion induced image degradation in high resolution peripheral computed tomography: Impact of image quality on measures of bone density and micro-architecture

Abstract

Motion artifacts are a common finding during high-resolution peripheral quantitative computed tomography (HR-pQCT) image acquisitions. To date it is not clear (i) when to repeat an acquisition, (ii) when to exclude a motion-degraded dataset post hoc, and (iii) how motion induced artifacts impact measures of trabecular and cortical parameters. In this study we present inter- and intra-observer reproducibility of a qualitative image quality grading score and report the prevalence of repeat acquisitions in our population. Finally the errors in bone density and micro-architectural parameters estimated from repeat acquisitions with and without motion degradation are presented. The relationship between these errors and the image quality grade is evaluated for each parameter. Repeat acquisitions performed due to operator-observed motion in the reconstructed image occurred for 22.7% of the exams (29.7% radius, 15.7% tibia). Of this subset, 88 exams with repeat acquisitions had at least one acquisition graded 1 (best quality). In this subset, the percent differences in bone density and micro-architecture measures tended to increase as the relative image quality decreased. Micro-architectural parameters were more sensitive to motion compared to geometric and densitometric parameters. These results provide estimates of the error in bone quality measures due to motion artifacts and provide an initial framework for developing standardized quality control criteria for cross-sectional and longitudinal HR-pQCT studies.