Information Technology Reference
In-Depth Information
High-resolution MRI and hr-pQCT measurements of bone microstructure at
peripheral sites revealed important information on the spine [
59
,
65
]. To give an
example, Ladinsky et al. [
63
] demonstrated that trabecular bone microstructure
quanti
ed with high-resolution MRI in postmenopausal women contributes to
vertebral deformity burden independent of areal vertebral BMD. However, Eckstein
et al. [
66
] reported a substantial heterogeneity of bone strength among clinically
relevant skeletal sites and that its loss in osteoporosis may not represent a strictly
systemic process. Therefore, the direct assessment of bone microstructure at the
spine is advantageous to assess vertebral fracture risk and evaluate treatment
response at this anatomical location.
MDCT is the only imaging technique for high-resolution bone imaging at the
spine in-vivo [
58
]. Clinical whole-body MDCT can achieve a maximal in-plane
spatial resolution of about 250
m
2
with an axial slice thickness of 500
×
250
μ
μ
m
[
67
]. Thus, MDCT systems do not have the suf
cient spatial resolution to reveal the
true trabecular bone microstructure. However,
trabecular bone microstructure
parameters and
CT
(micro-CT) or hr-pQCT as standard of reference showed high correlations and
predicted biomechanically determined bone strength equally well [
68
finite element models (FEMs) assessed with MDCT and
μ
-
70
]. Bauer
et al. harvested 20 cylindrical
trabecular bone specimens from formalin-
xed
μ
human thoracic spines [
68
].
CT images of the bone specimens were obtained with
an isotropic voxel size of 20
μ
m
3
and corresponding MDCT images up to a voxel
size of 230
m
3
. Trabecular bone microstructure parameters obtained
×
230
×
500
μ
CT and MDCT showed R
2
values up to 0.84. Furthermore, MDCT derived
trabecular bone microstructure parameters demonstrated high correlations with
biomechanically determined bone strength (R
2
values up to 0.81). Similarly, Baum
et al. [
69
] examined formalin-
from
μ
xed spinal segment units by using hr-pQCT (iso-
m
3
) and a clinical whole-body MDCT (spatial resolution of
tropic voxel size of 41
μ
m
3
). Corresponding images of a spinal segment unit acquired
with MDCT and hr-pQCT as standard of reference are shown in Fig.
9
. Correlations
between trabecular bone microstructure parameters and biomechanically deter-
mined failure load amounted up to r = 0.86 using the hr-pQCT images, and up to
r = 0.79 using the MDCT images. Correlation coef
250
×
250
×
600
μ
cients of failure load versus
trabecular bone microstructure parameters obtained with HR-pQCT and MDCT
were not signi
cantly different. Furthermore, no differences in the performance of
64- and 320-slice MDCT scanners with respect to the depiction of trabecular bone
microstructure were observed [
71
].
The calculation of bone microstructure parameters and FEMs in MDCT images
of the spine requires several steps including image registration and segmentation.
Multiple (semi-) automated image registration and segmentation algorithms have
been developed to minimize time effort and reproducibility errors, which are par-
ticularly important for the assessment of change in longitudinal studies [
72
-
76
].
Thereby, regions of interest (ROIs) are drawn in the acquired images to de
ne the
outer contour of the vertebrae and consecutively certain areas of the trabecular
bone.
