Evaluation of technical factors affecting the quantification of trabecular bone structure using magnetic resonance imaging.

High resolution magnetic resonance (MR) techniques combined with standard techniques of stereology and texture analysis have been used to quantify trabecular structure. Using dried excised specimens from the tibia (n = 10) and radius (n = 2) we evaluate the impact of using volumetric gradient-echo (GE) and spin-echo (SE) MR imaging sequences, the relative importance of echo time in gradient-echo MR imaging, and the impact of different threshold values to segment the bone and bone marrow on the estimation of trabecular bone structure. We also investigate the inter-relationships between the different structural parameters derived from MR images. Images were obtained using fast gradient-echo and spin-echo imaging sequences, with TE values ranging from 7 to 17 ms using 4.7 and 1.5 Tesla imaging systems. In-plane image resolution ranged from 128 to 156 microns, and slice thickness ranged from 128 to 1000 microns. We derived stereological measures such as the mean intercept length, trabecular width, fractional area of trabecular bone, trabecular number, and trabecular spacing, the fractal dimension as a texture-related parameter and the Euler number as a measure of connectivity from these images. We found that the mean intercept length as a function of angle traced an ellipse with the orientation of the principal axis of the ellipse, a measure of trabecular orientation, identical when measured from the spin-echo or gradient-echo MR images. Absolute measures such as the fractional area, trabecular width, trabecular number, and fractal dimension as measured from gradient echo images were 28%, 30%, 1.3%, and 0.6% greater, respectively, than those calculated from spin-echo images, while the trabecular spacing was 14% less when calculated from gradient-echo images compared to spin-echo images. The structural parameters also depended on the echo time used to obtain the MR image. The choice of the threshold used to segment the high resolution images also affected the estimated structural parameters significantly. Our results indicate that MR may be used to visualize and quantify trabecular bone architecture; however, the different technical factors that could affect the appearance of MR images must be understood and considered in the data analysis and interpretation.