Comparative investigation of bone mineral density using CT and DEXA in a canine femoral model

Abstract

Bone density measurements using computed tomography (CT) instead of dual-energy X-ray absorptiometry (DEXA) are currently of great interest in human and veterinary medical research as it would be beneficial to use CT scans obtained for other indications also for determining bone density. For Hounsfield units (HU) measured with CT in specific regions of interests (ROIs) in one or several slice/s a correlation with bone mineral density (BMD) measured by DEXA in humans and dogs of between 0.44 and 0.77 is reported in the literature. In the present study, instead certain volumes of interest (VOIs) obtained by CT scan and the corresponding HU to the respective VOIs were compared with the bone mineral density of the corresponding areas measured by DEXA. The aim of the study was to investigate whether this procedure gives more accurate information about bone density of the bones as three-dimensional objects of the respective patient. Correlation between measured HU in the respective VOI and BMD measured with DEXA in the corresponding ROI showed a very good correlation of 0.93. Linear regression with R = 0.85 (p = 0.0262) was calculated. Except for VOI5, similar distribution of values and significant differences (p < 0.0001-0.0087) between ROIs/VOIs were detected. Determining HU for assessing bone mineral density in a certain volume provides more accurate results than those previously reported from two-dimensional (2D) CT measurements. © 2017 Orthopaedic Research Society. Published by Wiley Periodicals, Inc. J Orthop Res 35:2667-2672, 2017.

Is bone mineral density lower in the necrotic lesion in pre‐collapse osteonecrosis of the femoral head?

“…HU values have been shown to correlate with BMD obtained using various measurement methods, such as DXA and quantitative CT 8‐11 . Evaluation of HU values can be performed using any slice of clinically obtained plain CT scans, which enables a detailed BMD assessment of various bone structures even though these cannot be evaluated by DXA 8,12,13 . To the best of our knowledge, no previous reports have investigated the HU values of the necrotic lesion in pre‐collapse ONFH.…”

“…7 HU values have been shown to correlate with BMD obtained using various measurement methods, such as DXA and quantitative CT. [8][9][10][11] Evaluation of HU values can be performed using any slice of clinically obtained plain CT scans, which enables a detailed BMD assessment of various bone structures even though these cannot be evaluated by DXA. 8,12,13 To the best of our knowledge, no previous reports have investigated the HU values of the necrotic lesion in pre-collapse ONFH. The purpose of this study was to verify whether HU values obtained from plain CT scans were useful for evaluating the BMD of femoral heads and to clarify whether the BMD of the necrotic lesion in pre-collapse ONFH was reduced.…”

Computed Tomography Osteodensitometry for Assessment of Bone Mineral Density of the Canine Head—Preliminary Results

“…Thus, CT scans have become an opportunity to obtain a global assessment of BMD. It is accepted that CT supersedes DEXA scans in veterinary patients when extreme variety of patient sizes and soft tissue dimensions render DEXA values inaccurate [ 45 ]. Veterinary publications that address qCT BMD are limited by small case numbers, use of cadavers, or use of a research population [ 14 , 40 , 46 , 47 , 48 , 49 ].…”

Computational tibial bone remodeling over a population after total knee arthroplasty: A comparative study

“…Firstly, the clinical scans included soft tissue structures around the bone, which contributed to the total projected density, while the projected FE densities were based on only the segmented tibiae, resulting in lower BMD as encountered in the proximal ROIs. And secondly, density differences could have been caused by the use of a different image modality in the two populations; although a very good correlation had been found between HU measured in CT and DEXA‐derived BMD, 52 and in the scan‐specific linear fit between HU and cubic BMD, 39 any deviations could be amplified over both established relations. Moreover, considerable speckle noise was encountered throughout CT images of the FE subjects, affecting the histogram‐based air–fat–muscle calibration, 39 in addition to the direct effect of speckle on voxel intensities; the calibration was performed using only the air and muscle values in clinical scans in which no explicit fat tissue intensity peak was observed due to the speckle, making the HU–BMD relation more prone to deviations.…”