Structural parameters of normal and osteoporotic human trabecular bone are affected differently by microCT image resolution

• Published in: Osteoporosis international 2011, Vol.22 (1), p.167-177
• Main Authors: Isaksson, H, Töyräs, J, Hakulinen, M, Aula, A. S, Tamminen, I, Julkunen, P, Kröger, H, Jurvelin, J. S
• Format: Article
• Language: English
• Subjects: Adult; Aged; Biological and medical sciences; Bones; computed tomography; Diseases of the osteoarticular system; Endocrinology; Female; Femur - diagnostic imaging; Humans; Ilium - diagnostic imaging; Image resolution; Knee Joint - diagnostic imaging; Male; Medical sciences; Medicine; Medicine & Public Health; Middle Aged; Original Article; Orthopedics; Osteoporosis; Osteoporosis - diagnostic imaging; Osteoporosis, Postmenopausal - diagnostic imaging; Osteoporosis. Osteomalacia. Paget disease; Radiographic Image Interpretation, Computer-Assisted - methods; Rheumatology; Thresholding; Tibia - diagnostic imaging; Tomography; Trabecular bone; X-Ray Microtomography - methods; Young Adult
• ISSN: 0937-941X; 1433-2965
• DOI: 10.1007/s00198-010-1219-0

Summary This study employed microCT to investigate whether image resolution affects bone structural parameters differently in healthy normal and osteoporotic trabecular bone. With increasing image voxel size, the originally detected differences between sample groups diminished. The results suggest that structural differences may not be reliably detected with clinical scanners. Introduction Structural parameters of bone reflect its health status, but are highly dependent on the image resolution. We hypothesized that image resolution affects bone structural parameters differently in normal and osteoporotic trabecular bone. Methods Human trabecular bone samples from the iliac crest and the knee were analyzed (normal n = 11, osteoporotic n = 15) using a high-resolution microCT (14 or 18 µm voxel sizes). Images were re-sampled to voxel sizes 1-16 times larger than the original image and thresholded with global or local adaptive algorithms. Absolute and normalized values of each structural parameter were calculated, and the effect of decreasing image resolution was compared between the normal and osteoporotic samples. Results Normal and osteoporotic samples had different (p < 0.05) absolute bone volume fractions. However, the normalized values showed that the osteoporotic samples were more prone to errors (p < 0.05) with increased voxel size. The absolute values of trabecular number, trabecular separation, degree of anisotropy, and structure model index were different between the groups at the original voxel size (p < 0.05), but at voxel sizes between 60 and 110 µm, those differences were no longer significant. Conclusions The results suggest that structural differences between osteoporotic and normal trabecular bone may not be reliably detected with clinical CT scanners providing image voxel sizes above 100 µm.