Quantification of bone microstructure in the wrist using cone-beam computed tomography.

Due to the rising life expectancy, bone diseases (e.g. osteoporosis, osteoarthritis) and trauma (e.g. fracture) have become an important socio-economic burden. Accurate visualization and quantification of the bone microstructure in vivo is seen as an important step to enhance diagnosis and treatment. Micro-computed tomography (microCT) has become the gold standard in three-dimensional (3D) imaging of trabecular bone structure. Yet, usage is limited to ex vivo analyses, hence, it cannot be used to evaluate bone and bone adaptive responses in a patient. High-resolution peripheral computed tomography (HR-pQCT) is considered the best technique to measure the bone microarchitecture in vivo. By design HR-pQCT is limited to scanning extremities, such as the distal radius and distal tibia with a limited field of view and long scanning time (~2 à 3 min. for a stack of 0.9 cm). Cone-beam computed tomography (CBCT) is a promising alternative with a much larger field of view. Yet, CBCT is challenged by artefacts that reduce image contrast, such that it is currently being used for qualitative evaluation only. Therefore, the aims of this work were first to enhance image contrast and second to determine the accuracy of high-resolution CBCT for bone microarchitectural assessment. Trapezia of nineteen female arthritic patients were scanned twice ex vivo; once using CBCT (NewTom 5G, Cefla, Verona, Italy) at a nominal voxel size of 75 μm and once using microCT (SkyScan 1172, Bruker, Kontich, Belgium) at a voxel size of 19.84 μm. The CBCT-scans were reconstructed following 2 protocols: (1) using the commercial software delivered with the scanner and (2) using in-house developed software. After reconstruction and image processing, the images were segmented using adaptive thresholding. Bone morphometric parameters including bone volume (BV), total tissue volume (TV), bone volume fraction (BV/TV), bone surface density (BS/TV), trabecular thickness (Tb.Th), trabecular separation (Tb.Sp) and trabecular number (Tb.N) were calculated. Statistical evaluations were made at a significance level of 5%. Significant correlations were found between the CBCT-based bone parameters and the microCT-based parameters with R > 0.68 The in-house reconstructed software outperformed the commercial software. Smaller bias (overestimation of Tb.Th decreased from 114.24% to 59.96%) as well as higher correlations were observed for the in-house processed images. Still, a significant overestimation was observed for BV/TV and Tb. Th and an underestimation for Tb.N. We conclude that our CBCT image reconstruction improved image contrast which allowed for an accurate quantification of trabecular bone microarchitecture.