HIGH-RESOLUTION MICRO-CT EXAMINATION OF HUMAN BONE: TOWARDS THE ENTIRE ORGAN
1
Egon Perilli, 2,3Ian H Parkinson, and 1Karen J Reynolds
1
Medical Device Research Institute, School of Computer Science, Engineering and Mathematics, Flinders University, SA
2
Bone and Joint Research Laboratory, SA Pathology and Hanson Institute, SA
3
Discipline of Anatomy and Pathology, The University of Adelaide, SA
email: [email protected]
INTRODUCTION
Micro-CT systems are available that facilitate ex vivo
examinations of human bones as large as entire vertebrae, with
spatial resolutions in the 10-micrometer range. Accurate 3Ddescription of the microarchitecture of entire organs can be
obtained, at resolutions previously achievable only on biopsies
[1,2]. These high-resolution scans produce large datasets, with
costs and benefits, which have to be considered.
The aim of this study is to present whole human vertebrae
scanned at high resolution (17 μm/pixel). The datasets were
down-sampled to 34 and 68 μm pixel size, their morphometric
parameters compared to those at 17 μm pixel size, and
discussed in relation to data size and calculation time.
METHODS
Five human L3 vertebrae were used, from 5 cadavers.
Posterior vertebral elements were removed. Scans were
performed with an in vivo micro-CT system (model 1076,
Skyscan NV, Kontich, Belgium), at 17.4 μm isotropic pixel
size [1]. For each vertebra, the full height (32 mm) was
reconstructed (68x68 mm cross-sections). The computer was
equipped with two dual-core Intel Xeon CPUs, each at 3.00
GHz, 4 GB memory, OS Windows XP 64 bit. Morphometric
parameters were calculated over the entire vertebra, using 3Dmethods [1]: trabecular bone volume fraction (BV/TV),
trabecular thickness (Tb.Th), trabecular separation (Tb.Sp),
structure model index (SMI), cortical thickness (Ct.Th).
Next, vertebral datasets were down-sampled by a factor of two
and four (34 and 68 μm/pixel), and the 3D-morphometric
parameters recalculated. To test for changes in morphometric
values at varying pixel size, ANOVA for repeated measures
was performed. Where statistical significance was found, a
paired t-test between morphometric data at lower resolutions
(34 and 68 μm/pixel) and highest resolution (17 μm/pixel) was
performed. Significance level was set to p = 0.05.
RESULTS AND DISCUSSION
High-resolution imaging of entire vertebrae produced large
datasets (26 GB, Tab.1). Down-sampling of the datasets did
substantially reduce calculation times for 3D-analysis, from
days to minutes (Tab.1).
Table1: Data size and calculation times at varying pixel sizes, per vertebra.
Pixel size:
17 µm
34 µm
68 µm
Entire vertebra, dataset size:
26.4 GB 3.3 GB
423 MB
3D morphometry, calculation times:
5 days
114 min
5 min
The 3D-morphometric parameters for both trabecular and
cortical bone showed significant monotonic changes at
varying pixel size, compared to 17 μm/pixel data (p < 0.05)
(Fig.1). The biggest changes were found for Ct.Th, Tb.Th, and
Tb.N (+56%, +53%, and -45%, at 68 μm/pixel).
(a)
(b)
(c)
(d)
Figure 1: Left: Percentage changes of 3D-morphometric parameters
at varying pixel size, with reference to 17μm/pixel. Right: (a) MicroCT 3D-rendering of a vertebral body, (b) trabecular bone, (c) cortical
bone, (d) virtual cylindrical trabecular bone biopsy.
CONCLUSIONS
As shown in this study, high-resolution micro-CT imaging of
entire vertebrae produces large datasets, challenging for 3Danalysis. Down-sampling introduced losses in accuracy in
morphometric parameters, as expected [3]. Nonetheless,
down-sampling was very effective in speeding up calculations.
One might ask whether it is really necessary to use a highresolution micro-CT scan (17 μm/pixel), if data might later be
down-sampled, for practicality [1]. The answer depends on the
research question, the minimum acceptable error for the study
purpose, and on computational resources available.
High-resolution micro-CT scans of entire bones give freedom
of choice: regions of interest can be extracted ad hoc, ranging
from the whole organ to sub-volumes of various sizes and
shapes, virtual biopsies (Fig.1, right). High-resolution scans
ensure higher accuracy even after down-sampling, compared
to low-resolution scans with similar large pixel size [3]. This
confirms high-resolution micro-CT as an invaluable tool for
bone investigations in 3D.
REFERENCES
1. Perilli E, et al., Ann Ist Super Sanita. 48:75-82, 2012.
2. Perilli E, et al., Bone. 50:1416-25, 2012.
3. Kim DG, et al., Bone. 35: 1375-82, 2004.