Reliability of tibial measurement with mechanical response tissue analysis.

摘要

Mechanical response tissue analysis (MRTA) provides a noninvasive means of estimating the cross-sectional bending stiffness (EI) of long bones, and thus can serve as a predictor of bone strength. Estimates of bone bending stiffness are derived from the point impedance response of a long bone to low frequency (70--500Hz) stimulation according to beam vibration theory. MRTA has demonstrated the ability to reliably estimate human ulnar bending stiffness with between-test coefficients of variation of 5%, and in vivo measurements of monkey tibiae have been validated with ex vivo 3-point mechanical bending tests. Human tibial MRTA measurement has achieved between-trial coefficients of variation of only 12%, so a new physical MRTA configuration and improved computer algorithms have been developed in an attempt to improve upon this level of reliability. The new configuration removes the rigid proximal and distal tibial restraints and models the tissue behavior with a 12-parameter algorithm that accounts for free vibration at the ankle and knee joints. Initial testing with only the hardware changes and application of the 7-parameter model of tissue behavior used in earlier systems yielded unacceptable variation. Subsequent reliability testing with application of 6-, 9-, and 12-parameter models demonstrated modest improvements, prompting the development of the more robust 12-parameter model used in the present study. Evaluation of 110 college-age females (age 20.2 +/- 1.8 yr, height 163.3 +/- 5.9 cm, weight 60.7 +/- 9.3 kg, BMI 22.8 +/- 3.1 kg·m-2) with the current MRTA system has demonstrated an improvement in within-trial reliability for unsupported tibial El measurement with a coefficient of variation of 11.2%. These results demonstrate the ability of the system to measure tibial response characteristics when both proximal and distal ends are free of rigid support. Long-term measurement reliability is still problematic with a coefficient of variation of 36.5% for a set of 4 measurements spanning 21 months.