Technical note: Does scan resolution or downsampling impact the analysis of trabecular bone architecture?

Abstract

The "gold standard" for the assessment of trabecular bone structure is high-resolution micro-CT. In this technical note, we test the influence of initial scan resolution and post hoc downsampling on the quantitative and qualitative analysis of trabecular bone in a Gorilla tibia. We analyzed trabecular morphology in the right distal tibia of one Gorilla gorilla individual to investigate the impact of variation in voxel size on measured trabecular variables. For each version of the micro-CT volume, trabecular bone was segmented using the medical image analysis method. Holistic morphometric analysis was then used to analyze bone volume (BV/TV), anisotropy (DA), trabecular thickness (Tb.Th), spacing (Tb.Sp), and number (Tb.N). Increasing voxel size during initial scanning was found to have a strong impact on DA and Tb.Th measures, while BV/TV, Tb.Sp, and Tb.N were found to be less sensitive to variations in initial scan resolution. All tested parameters were not substantially influenced by downsampling up to 90 μm resolution. Color maps of BV/TV and DA also retained their distribution up to 90 μm. This study is the first to examine the effect of variation in micro-CT voxel size on the analysis of trabecular bone structure using whole epiphysis approaches. Our results indicate that microstructural variables may be measured for most trabecular parameters up to a voxel size of 90 μm for both scan and downsampled resolutions. Moreover, if only BV/TV, Tb.Sp or Tb.N is measured, even larger voxel sizes might be used without substantially affecting the results.