Biomedical Engineering Reference
In-Depth Information
computed tomography (QCT) images (Müller and Rüegsegger,
1996
). In the study,
3D FE models were paired with controlled Gaussian filtration to derive two models
of moderate and pronounced bone atrophy. While the resulting apparent Young's
moduli were the only mode of validation for this early attempt of
in silico
simula-
tion of bone adaptation, the reported methods became the foundation for subsequent
models. Thus, a follow-up study introduced further improvement to the application
by using 3D micro CT scans as the input for the algorithm (Müller and Hayes,
1997
). The isotropic resolution of the input images increased from 170 to 14 µm. In
addition, the original simulated bone atrophy (SIBA) model was expanded to fol-
low the mechanostat hypothesis (Frost,
1964
), allowing controlled formation and
resorption. This, in turn, facilitated simulation of various stages of bone loss, as
well as a more realistic 'age-match'. For validation the results were compared to
the experimental measurement of the post-menopausal group both qualitatively and
quantitatively, and proved to be in good agreement. Finally, the model was applied
to the 3D micro CT scans of human iliac crest and lumbar spine biopsies selected
from the pre- and post-menopausal groups, in an attempt to simulate pre-, peri-,
and post-menopausal bone states (Müller,
2005
). In this study strong emphasis was
placed on the validation of the results against biological data. Thus, visual compar-
ison after the simulation of 43 years confirmed that the model produced realistic
trabecular architecture when compared to the
in vivo
group, while quantitative bone
morphometry, carried out for both groups produced a 100 % match for the bone vol-
ume density (BV/TV) parameter, and excellent agreement for the other parameters.
Another
in silico
simulation, based on true bone geometry and verified against
in vivo
biological data, employed a voxel-based surface adaptation under uniaxial
compression (Adachi et al.,
2001
). In this algorithm, micro-CT measurements of
canine cancellous bone were obtained from the previously published investigation
(Goldstein et al.,
1991
), and the results of the simulation were compared to the
corresponding animals at the end of the
in vivo
study (Guldberg et al.,
1997
). The
validation was performed based on a comparison of the calculated morphometric in-
dices for the
in silico
and
in vivo
experiments respectively. The results from the two
approaches were in good agreement with values for bone volume fraction (BV/TV)
being 0
.
230 and 0
.
222 for the experimental and simulated samples, respectively.
Several other notable studies, presenting elaborate models with realistic results,
should be mentioned. For example, long-term investigation performed on the 3D
micro-CT scans of human vertebra modeled the period of 50 years (Van der Lin-
den et al.,
2001
). Morphometric indices, calculated for the resulting structures, cor-
related closely with the values reported in literature. Unfortunately, no validation
against experimental data has been performed for this study. Another remarkable
algorithm has been presented by Ruimerman et al. (
2005a
). The simulation was car-
ried out on computer generated cubes of trabecular bone, and investigated the abil-
ity of bone to adapt in response to elevated strains. In addition to implementing the
theory for metabolic expression under load (Huiskes et al.,
2000
), an extensive ex-
amination of osteocytic stimuli, such as maximal principal strain, volumetric strain,
and strain energy density (SED) have been carried out as part of the study (Ruimer-
man et al.,
2005b
). While the model parameters themselves are largely based on the
