Microstructural quality of vertebral trabecular bone can be assessed from ultrasonic wave propagation

Abstract

Ultrasound is extensively studied as an alternative diagnostic "screening" tool for osteoporosis. However, only a few aspects about the interaction of ultrasonic waves with bone tissue are understood, primarily based on statistical correlations. Therefore, our long-term aim is to obtain a more mechanistic interpretation of ultrasonic wave propagation through bone microstructure. For this study our aim was to quantify the interplay between ultrasound frequency and bone microstructure in determining wave propagation. Three-dimensional (3D) numerical simulations were performed on 15 human trabecular bone samples from the lumbar spine (4×4×4 mm 3). The 3D representation of trabecular bone architecture was obtained using microcomputed tomography at a resolution of 14 micrometer. The simulations were performed using commercial finite element (FE) software (MSC/Nastran), validated for this application. Three frequencies (1 MHz, 300 kHz and 50 kHz) were analyzed, and the velocity of wave propagation (Speed of Sound, SOS) was calculated. SOS showed no correlation with bone volume fraction (BV/TV) and a better correlation with the apparent elastic modulus. The variables were calculated in the three main directions of the bone sample. Our simulations at 50 kHz and 300 kHz frequency showed an excellent agreement between FE-calculated SOS and the SOS estimated with the bar velocity equation (R2 = 0.93 and R2 = 0.85 respectively). For 1 MHz, this correlation was significantly lower (R2 = 0.66), but could be substantially improved by including several morphometric parameters in a multiple regression (R2 = 0.79). These results show that for low frequencies (50 kHz-300 kHz) of the input wave, the effects of bone micro-structure can be assessed at the macroscopic level. For higher frequencies (1 MHz) the results deviated more from the bar equation. Interestingly, the remaining variation could be largely explained by microstructural parameters. Hence, we hypothesize that ultrasound analyses can yield information on the microstructural and on macrostructural properties when performed with several frequencies. © 2009 Springer Berlin Heidelberg.