Calculating micro- and macro-scale bone quality parameters from computed tomography images using a parametric physical model and standard bone composition

Abstract

Bone quality parameters (BQPs) calculated from computed tomography (CT) and micro-computed tomography (μCT) images have been applied in clinical to diagnose and prevent osteoporotic fractures. However, the BQPs calculated from CT and μCT images are at different scales and incomparable due to the differences in image resolutions and accounted bone compositions. In this study, we proposed a novel method to convert HU values of CT images into the BQPs that were originally calculated using different tools. By using the standard bone composition of human bones, the BQPs of bone volume fractions (BV/TV) and volumetric bone mineral densities (vBMD) of the bones were calculated using different accounted compositions of bones and mathematical relationships defined in this study. In addition, the HU values of the bones were estimated using a parametric physical model and spectrum characteristic parameters. Relationships for converting HU values into BQPs were determined by linear fitting the calculated BQPs and estimated HU values. Cadaveric and patient studies show that the micro-scale bone volumes and macro-scale vBMDs of vertebrae calculated from CT images using the conversion relationships determined by the proposed method were close to those calculated from the μCT images and quantitative CT, respectively, with mean percent errors lower than 5%. The proposed method could be applied in clinical to accurately compute micro- and macro-scale BQPs from CT images for assessing the bone quality of patients.