The ribs play a key structural role in the chest during dynamic loading, and future safety advancements will increasingly rely on the biofidelity of Human Body Models (HBMs). Rib geometry in HBMs is typically obtained via global imaging from CT, however for thin cortices such as those in ribs, traditional histogram-based segmentation is known to produce unacceptably high error. Here we test the accuracy of a new cortical bone mapping (CBM) algorithm in resolving these local geometries from globally obtained full-body CT scans, as well as from higher resolution scans of isolated ribs. Results from 11 cadaveric subjects' scanned ribs are compared to ground truth histology obtained from specific rib sections.

The CBM methodology resolved the local thickness of cortical bone from isolated rib scans to within 0.01±0.14mm (mean±SD) error, and from full body scans to -0.04±0.20mm error. Average positional error in the placement of the rib's periosteal border was -0.01±0.13mm from full body scans using CBM, compared to 0.81±0.42mm for histogram-based thresholding of those same scans. Consequently, important mechanical properties are resolved to much higher accuracy using CBM, with total rib cross sectional area error reduced from 39±12% under traditional thresholding to under 5±2% when obtained using CBM.

This study has shown that – given appropriate extraction methodologies – local geometric properties of rib bones can be obtained from full-body CT imaging of human subjects. Results from this study open the door for subject-specific rib cortical bone thickness maps to be obtained for HBM validation and improvement.