Information Technology Reference
In-Depth Information
Fig. 12
1
H-MRS of L4 at 3T: Sagittal (a) and axial (b)T
2
T2-weighted images with the red box in L4
indicating the position for the employed single-voxel
1
H-MRS. The
1
H-MRS-based spectrum of
L4 (c) shows the methyl (
(CH
2
)
n
CH
3
) peak at 0.9 ppm (peak 1), the superposition of the
-
-
methylene (
(CH
2
)
n
) peak at 1.30 ppm and the
ʲ
-carboxyl (
CO
CH
2
CH
2
) peak at 1.59 ppm
-
-
-
-
-
-
(peak 2), the superposition of the
ʱ
-ole
nic (
CH
2
CH=CH
CH
2
) peak at 2.00 ppm and the
-
-
-
-
ʱ
-carboxyl (
CO
CH
2
CH
2
) peak at 2.25 ppm (peak 3), the water peak at 4.7 ppm (peak 4), and
-
-
-
-
the ole
nic (
CH=CH
) peak at 5.3 ppm (peak 5)
-
-
T
1
T1-weighted MR images expressed as coef
cient of variation (CV) amounted 1.0
and 2.6 %, respectively [
101
]. The main error source for the calculation of bone
marrow adipose tissue volume in T
1
T1-weighted MR images results from partial
volume effects and the threshold selection, making the technique semi-quantitative
[
101
]. Chemical shift-based water-fat separation techniques extract the bone marrow
fat fraction based on the different precession frequencies of water and lipid hydrogen
protons. Confounders such as T
2
* decay, T
1
effects, the multi-spectral nature of fat,
and noise bias have to be addressed for reliable fat quanti
109
]. After
the correction of these confounding factors, the proton density fat fraction (PDFF)
can be quanti
cation [
107
-
ed [
110
]. Regions of interest (ROIs) have to be placed in vertebral
body, e.g. by using the
first in-phase series in mid-sagittal view of the vertebral
bodies [
107
]. Then, the ROIs are copied onto the reconstructed PDFF map to
determine the vertebral bone marrow fat fraction [
107
].