Search

An important feature of humeral orthopedic finite element (FE) models is the trabecular stiffness relationship. These relationships depend on the anatomic site from which they are derived; but have not been developed for the humerus. As a consequence, humeral FE modeling relies on relationships for other anatomic sites. The variation in humeral FE outcomes due to the trabecular stiffness relationship is assessed. Stemless arthroplasty FE models were constructed from CT scans of eight humeri. Models were loaded corresponding to 45 deg and 75 deg abduction. Each bone was modeled five times with the only variable being the trabecular stiffness relationship: four derived from different anatomic-sites and one pooled across sites. The FE outcome measures assessed were implant-bone contact percentage, von Mises of the change in stress, and bone response potential. The variance attributed to the selection of the trabecular stiffness relationship was quantified as the standard deviation existing between models of different trabecular stiffness. Overall, variability due to changing the trabecular stiffness relationship was low for all humeral FE outcome measures assessed. The variability was highest within the stress and bone formation potential outcome measures of the trabecular region. Variability only exceeded 10% in the trabecular stress change within two of the eight slices evaluated. In conclusion, the low variations attributable to the selection of a trabecular stiffness relationship based on anatomic-site suggest that FE models constructed for shoulder arthroplasty can utilize an inhomogeneous site-pooled trabecular relationship without inducing marked variability in the assessed outcome measures. [ABSTRACT FROM AUTHOR]

Quantitative computed tomography (qCT) relies on calibrated bone mineral density data. If a calibration phantom is absent from the CT scan, post hoc calibration becomes necessary. Scanning a calibration phantom after-the-fact and applying that calibration to uncalibrated scans has been used previously. Alternatively, the estimated density is known to vary with CT settings, suggesting that it may be possible to predict the calibration terms using CT settings. This study compares a novel CT setting regression method for post hoc calibration to standard and post hoc phantom-only calibrations. Five cadaveric upper limbs were scanned at 11 combinations of peak tube voltage and current (80-140 kV and 100-300 mA) with two calibration phantoms. Density calibrations were performed for the cadaver scans, and scans of the phantoms alone. Stepwise linear regression determined if the calibration equation terms were predictable using peak tube voltage and current. Peak tube voltage, but not current, was significantly correlated with regression calibration terms. Calibration equation slope was significantly related to the type of phantom (p < 0.001), calibration method (p = 0.026), and peak tube voltage (p < 0.001), but not current (p = 1.000). The calibration equation vertical intercept was significantly related to the type of phantom (p < 0.001), and peak tube voltage (p = 0.006), but not calibration method (p = 0.682), or current (p = 0.822). Accordingly, regression can correlate peak tube voltage with density calibration terms. Suggesting that, while standard qCT calibration is preferable, regression calibration may be an acceptable post hoc method when necessary. [ABSTRACT FROM AUTHOR]