Independent Measurement of Femoral Cortical Thickness and Cortical Bone Density using Clinical QCT
Category: Biomechanics and Bone Quality
Poster Sessions, Presentation Number: SU0028
Session: Poster Session II & Poster Tours
Sunday, September 14, 2014 12:30 PM - 2:30 PM, George R. Brown Convention Center, Discovery Hall-Hall E
* , UNITED KINGDOM, Andrew Gee, University of Cambridge
Purpose: There is growing evidence that local features of cortical bone are a contributor to fracture risk. Both drug treatment and exercise regimes also result in cortical changes which are focused in particular regions rather than dispersed over the whole proximal femur. Accurate measurement of local cortical parameters is hence of importance in assessing both risk and intervention strategies. We have previously noted that both thresholding and full-width half-maximum (FWHM) techniques are incapable of measuring thickness or density for cortices less than 3 mm thick. Cortical bone mapping (CBM v1) can produce a much more accurate assessment of cortical thickness and mass (per unit cortical surface area), based on an estimate of cortical density which is presumed to be constant for each hip. However, although density varies much less than thickness, we would also like to be able to estimate the variation of density over the cortex.
Methods: A new CBM technique (v2) has been developed which makes use of the locally measured imaging blur to correct the initially global density estimate. This blur is overestimated when the initial density is too large, and underestimated when it is too small. The resulting local re-estimate of density can also be used to improve cortical thickness measurement. Performance was assessed with 70 ex vivo scans of proximal femurs (two each from 18 males and 17 females, aged from 56 to 96) from a study ethically approved by the Medical University of Vienna. True values were derived from calibrated HRpQCT data (cubic voxels of 0.082 mm) and estimates were made from QCT scans (0.33x0.33x1.0 mm) of the same femurs, each containing a BDC calibration phantom. Cortical thickness and density estimates were derived from over 700,000 matching measurements.
Results: Figure 1 shows the measurement errors for the FWHM, original CBM v1 and novel CBM v2 techniques. For thickness (in mm), range 1 to 3 mm, bias±std was 0.48±0.37 (FWHM), -0.24±0.32 (CBM v1) and 0.12±0.39 (CBM v2), and for 0.3 to 1 mm was 1.10±0.37 (FWHM), -0.24±0.12 (CBM v1) and -0.15±0.23 (CBM v2). Density estimation (in mg/cm3
) was less precise, but for range 1 to 3 mm cortex, results were -170±137 (FWHM), 195±217 (CBM v1) and -26±178 (CBM v2). Cortical density estimation for CBM v2 was relatively unbiased above 800 mg/cm3
Conclusion: It is possible to estimate both cortical density and thickness from clinical QCT, locally over the femur, though density estimation is less precise.
G. Treece, Eli Lilly & Co.: Research Grants. G. Treece, Amgen Inc.: Research Grants. A. Gee, Amgen Inc.: Research Grants. A. Gee, Eli Lilly & Co.: Research Grants.
* Presenting Authors(s):
, UNITED KINGDOM