Bone "quality" refers largely to the micro-geometry of trabecular networks (~10 - 200 mm length scale), but has been less well studied because of the need for (invasive) bone slices from biopsies. Studies often compare the topology of cancellous networks (eg trabecular number, number of nodes) with clinical outcomes such as fracture. It is desirable to obtain this information in-vivo from fracture-prone regions of the human body. However, invasive biopsies are routinely available from only one anatomical region, the iliac crest - which is not particularly susceptible to fracture . Cancellous bone within the heel bone, vertebral column or distal forearm provides excellent examples of trabecular networks that reflect the functional demands placed on them. Though it would be useful to image these regions in-vivo, non-invasive in-vivo imaging of trabecular networks at a resolution sufficient to distinguish individual trabeculae is currently only barely possible in selected regions of the appendicular skeleton, using MRI (Majumdar et al., Euro J Radiol 4:517, 1994). Thus while MRI may become the imaging modality of choice for such studies, currently it is not a feasible option.
However, it has long been recognised (eg Geraets & Van der Stelt, Pat Rec Letts 12:575, 1991) that information related to trabecular networks can be obtained from plain projection-radiography images of whole bones, since there is (at least) a qualitative relationship between the sizes and orientations of the image “elements” and those of the trabeculae generating them by superimposition. Studies in the Department of Medical Technology & Physics of Sir Charles Gairdner Hospital are examining the quality and clinical relevance of micro-structural information, obtained using quantitative image analysis , from both ex-vivo and in-vivo radiographs.
Digitised (256x256x8-bit) images of trabecular regions are analysed using Fourier (FT), wavelet transforms, and fractal dimensions. The general orientations (or sizes) of periodic elements identified in a 2-D Fourier “power” spectrum are determined from division of the Fourier plane into contiguous sectors (or annuli).
This project will be partly an introduction to image analysis software, as well as a means of obtaining interesting, clinically relevant results.