Image Processing Reference
In-Depth Information
is performed over the fixed frequency range, wrong resonance areas will be
computed. This can be overcome by using spectra-fitting algorithms that take
spectra shifts into account.
In any case, for reliable analysis of CSI data, an inspection of spectra in
considered areas is unavoidable.
11.3.7
C
OMPARISON
OF
SVS
AND
CSI T
ECHNIQUES
CSI offers the possibility of acquiring more spectra in the same time instead
of just one spectrum as with SVS, but does not bring about any improvement
in sensitivity. In terms of SNR per unit time and considering that the application
of phase encoding is equivalent to spectra averaging, the CSI and SVS tech-
niques are equally efficient (55,56). It should be pointed out that the equivalence
of both methods, in terms of sensitivity, assumes that SVS and CSI voxels have
the same size. As shown earlier, due to PSF effects, the area contributing to
the signal of the selected voxel extends the nominal voxel size. This fact leads
to the decreased sensitivity of the CSI experiment. This can be demonstrated
by the integration of PSF over the nominal voxel size (57), which in the case
of no filtering equals
∆
x
/
2
∆
x
/
2
sin(
π
Nx FOV
′
/
)
∫
∆
∫
PSF x
()
′
dx
′ =
x
xFOV
dx
x
x
′ ≈
0 873
.
(11.25)
N
sin(
π
′
/
)
−
x
/
2
−
∆
x
/
2
x
The loss of the voxel signal compared to an ideal selective excitation method
is apparent. The signal is not actually lost, but distributed among other voxels
(signal bleeding). The overall signal measured in each voxel depends, therefore,
on the detailed distribution of the measured signal in the tissue. This fact limits
the equivalence of both methods.
The preference of one or the other method depends on the number of averages
required to acquire spectra of the desired quality. In 2-D CSI using 16
16
matrix, signals from 256 voxels are measured in the same time and with the same
SNR as opposed to one voxel in SVS using 256 averages. In this case, CSI
becomes more efficient in terms of acquired information per unit time. On the
other hand, if only few averages are needed, e.g., for a reference spectrum without
water suppression, SVS is preferable.
In practical situations there are also other effects influencing the quality of
acquired spectra. The need for shimming of the large volume in the case of CSI
often means worse magnetic field homogeneity, broader signals, and, hence,
lower SNR.
On the other hand, measurements with smaller voxel size are often performed
with CSI. This leads to better spatial resolution and to improved magnetic field
homogeneity within the voxel, resulting in the significant decrease of the widths
×
Search WWH ::
Custom Search