Original Full Length ArticleChallenges in longitudinal measurements with HR-pQCT: Evaluation of a 3D registration method to improve bone microarchitecture and strength measurement reproducibility
Introduction
With the development of high-resolution peripheral quantitative computed tomography (HR-pQCT), a direct quantification of three important determinants of bone quality: microarchitecture, mineralization and biomechanical properties has become possible. This technique enables in vivo measurements at peripheral sites such as the distal radius and tibia, with a nominal isotropic voxel size of 82 μm. Bone microarchitecture parameters measured by HR-pQCT have been found to be associated with prevalent fracture in postmenopausal women and older men independently of areal bone mineral density (aBMD) [1], [2], [3], [4], [5]. Bone strength, estimated using micro-finite element analysis (μFEA), was also associated with prevalent fracture [6], [7], [8], [9], [10], [11], [12], [13], [14]. Moreover, HR-pQCT has been used to assess age-related bone loss [15], [16], [17], [18], [19] and to monitor variations in microarchitecture parameters during osteoporosis treatments [20], [21], [22], [23], [24], [25], [26].
The reproducibility of in vivo measurements is affected by significant sources of error including: poor calibration, movement artifacts, physical positioning, and scanning region selection for maximizing repeated measurements. By providing careful instructions to the subject, it is possible to improve reproducibility via reduction of movement artifacts, however the repositioning remains challenging.
With the standard scanning approach, immobilization of the arm or leg in an anatomically formed carbon fiber shell is used to provide support during the examination, and a reference line relative to anatomical landmarks is used to select the scanning region. Nevertheless, the repositioning of the limb in the exact same way remains a manual process with limited precision, and subsequently differences in scanning region exist between baseline and follow-up scans.
To overcome part of the repositioning error, a cross-sectional area (CSA) registration method was embedded within the HR-pQCT software (Image Processing Language, v5.16, Scanco Medical AG, Brüttisellen, Switzerland). This method was designed to correct axial misplacement between successive scans but did not take into account possible tilt of the limb. Given the dimensions of the scan region, even small tilt angles can lead to a considerable mismatch in the selected region. Only a three dimensional (3D) image registration process could provide an accurate alignment of the repeated scans, and thus may offer an improvement in the reproducibility of bone measurements by decreasing the repositioning error.
Indeed, Macneil et al. have reported an improved reproducibility with 3D registration based on gray-level images, most markedly for total and cortical volumetric BMD but not for trabecular morphological parameters [27]. Shi et al. introduced a 3D registration method based on registered masks after down-sampling gray-level images and demonstrated that bone density and trabecular architecture were comparable with those obtained using the CSA registration method [28].
So far, however, no study has reported the impact of 3D registration on the reproducibility of cortical microstructural parameters, which may be affected by small tilt errors more than the trabecular bone parameters. Also, none of these earlier studies investigated whether 3D registration could improve the reproducibility of the biomechanical parameters. Finally, the different studies conducted so far used different registration software and algorithms. Recently, a 3D registration algorithm has become available as part of the standard HR-pQCT software, which makes this available to all users. To date, no results were reported with regard to the reproducibility of this algorithm and implementation and validation relative to other studies was lacking.
The aim of our study was to apply a fully automated 3D registration process between repeated measurements to determine whether short-term reproducibility of bone microarchitectural and biomechanical parameters could be improved compared to using non-registered images and using the standard CSA-based registration method, at the radius and tibia.
Section snippets
Subjects
In vivo measurements were performed in 15 healthy subjects (aged 21–47 years), with no history of previous fracture, or bone-related ailments (recorded by a questionnaire). All of them were scanned three times within one month at both the non-dominant radius and tibia [29]. The time lag between two measurements varied between the same day, after complete repositioning, and a week. All measurements were performed by a single operator, who had been managing the system for one year.
The protocol was
Image quality
Among the 90 measurements, only one scan (grade 4) was excluded and repeated on the same day. The image quality ranged from grade 1 to grade 3 (grade 1: n = 10 / 25, grade 2: n = 22 / 17, and grade 3: n = 13 / 3, for the radius/tibia respectively).
3D registration accuracy
The 3D registration process was applied on a subset of 10 original and transformed images. The common VOI was thereafter applied to the original image with a mean overlap of 99.8% at the radius and 99.9% at the tibia compared to baseline VOI. Results between
Discussion
We have evaluated the in vivo clinical performance of the HR-pQCT system to longitudinally assess bone changes, by determining short-term reproducibility of geometry, density, microarchitecture and biomechanical measures at the distal radius and tibia, using two different registration methods.
We found that the CSA registration improved reproducibility of the geometry measures and the radius total density, compared with no registration. 3D registration further improved the reproducibility of the
Conflict of interest statement
Dr. van Rietbergen serves as consultant for Scanco Medical AG.
All other authors have no conflicts of interest.
References (51)
- et al.
Delmas PD, et al
Finite element analysis performed on radius and tibia HR-pQCT images and fragility fractures at all sites in postmenopausal women. Bone
(2010) - et al.
Regional, age and gender differences in architectural measures of bone quality and their correlation to bone mechanical competence in the human radius of an elderly population
Bone
(2009) - et al.
Longitudinal evaluation of the effects of alendronate on MRI bone microarchitecture in postmenopausal osteopenic women
Bone
(2011) - et al.
Improved reproducibility of high-resolution peripheral quantitative computed tomography for measurement of bone quality
Med Eng Phys
(2008) - et al.
Visual grading of motion induced image degradation in high resolution peripheral computed tomography: impact of image quality on measures of bone density and micro-architecture
Bone
(2012) - et al.
A three-dimensional digital image correlation technique for strain measurements in microstructures
J Biomech
(2004) - et al.
Reproducibility of bone micro-architecture measurements in rodents by in vivo micro-computed tomography is maximized with three-dimensional image registration
Bone
(2010) - et al.
A new method to determine trabecular bone elastic properties and loading using micromechanical finite-element models
J Biomech
(1995) - et al.
The elastic properties of trabecular and cortical bone tissues are similar: results from two microscopic measurement techniques
J Biomech
(1999) - et al.
A generalized least significant change for individuals measured on different DXA systems
J Clin Densitom
(2007)