Threshold-Free Automatic Detection of Cortical Bone Geometry by Peripheral Quantitative Computed Tomography

An accurate assessment of bone strength is an important goal in clinical bone research. For appropriate information on bone strength, precise segmentation of actual cross-sectional bone geometry is needed. In this article, we introduce an automatic, simple, and fast approach for reliable segmentation of cortical bone cross-sectional area based on the outer boundary detection and subsequent shrinking (OBS) procedure. Using repeated in vivo peripheral quantitative computed tomography (pQCT) images of distal tibia from 25 subjects, we compared new segmentation results with those obtained from commonly applied simple density thresholds and from a recent advanced analysis based on distance regularized level set evolution (DRLSE). Manual segmentation of cortical bone done by 3 independent evaluators was considered a gold standard. The new approach showed nearly 50% less variation in error compared with threshold-based analysis in conjunction with a recently introduced statistical preprocessing method and agreed well with results obtained from manual segmentation. The DRLSE segmentation resulted consistently in ∼15% mean overestimation of all geometrical traits with a similar variation of data as obtained from the OBS method. In conclusion, the OBS method improved assessment of all observed measures of cortical geometry and can enhance the cortical bone analysis of pQCT images in clinical research studies.