Repositioning of the region of interest in the radius of the growing child in follow-up measurements by pQCT

Abstract

The assessment of changes in bone mineral density (BMD) requires follow-up measurements. In order to compute accurate changes, it is important that the region of interest (ROI) of the initial and follow-up measurements match in terms of location. Our present focus is the evaluation of bone at the distal end of the radius by peripheral quantitative computed tomography (pQCT). In adults, in whom the bones have ceased to grow, repositioning of the ROI has been attempted either by matching the measurement location defined as a fixed percentage of the bone length away from an anatomical landmark or by matching the cross-sectional areas of the bone in the conic distal region of the radius; however, in the case of children, whose bones are growing, these methods cannot be applied blindly. The aim of this study was to propose a model that aids in the relocation of the ROI during follow-up measurements in children. From a long-term research study of multiple generations of volunteers, we were able to obtain sets of radiographs of forearms taken regularly from birth until the age of 18 years. From the radiographs of each subject, the bone lengths and widths were measured. The shapes of the bones were analyzed, and a model that predicts change in width based on change in length was developed. This model was modified to predict change in cross-sectional area based on change in length, assuming a cross-sectional geometry derived from a set of pQCT images over the same age range. It was observed that the shape of the radiographically projected radius between distal and proximal growth plates for a given subject did not vary with age. The model was tested by matching pairs of pQCT scans based on the cross-sectional areas predicted by the model. The pairs of images were compared qualitatively for a match in shape of the cross-section of the bone. If the cross-sectional shape matched closely, the two sections were considered to belong to the same region. From 26 evaluated pairs of cross-sections, 21 pairs had matching shapes. From this study we concluded that the proposed model produced satisfactory results in 81% of the tested cases. The failures were observed mostly at an age when the increase in length ceased but the increase in cross-section continued. The results of the qualitative evaluation is deemed encouraging and warrants further tests to see if ROIs defined with this model provide clinically meaningful results. © 2007 Springer-Verlag Berlin Heidelberg.