Carte RMN PDF
Carte RMN PDF
Carte RMN PDF
k-Space Formalism
o 0 0 0 0 0 0
o 0 0 0 0 0 0
0 0 0 0 0 0 0
E
0
0
0
0
0
0
0
0
0
0
0
0
0
0
--
0
LO
0
o 0 0 0 DOD
o 0 0 0 DOD
1 / cm
spatial
frequency
Figure A.l. The spatial frequencies of an image comprising a matrix of
dots.
Image K-space
Space
.. .
,
/
I I
!
/
I ...
::-.l \
• ,1
,
I I ,.
N :~ <J :>-
I
.. ' ,
:' 1 •
': / Fourier
\ ' I!~ ...
; 11J/1
Transfor
"1\ I
I ;·f~· \' ~ ;
. ,.I:
"\
I
" 'II
.\
the link between Fourier space and images is direct, since data
are collected in the computer memory in terms of spatial
frequencies.
Phase Encoding
The k-space image is directly connected to the MRI image by the
Fourier transform operation. As described in Chapter 2,
the phase-encode technique provides an elegant way to encode
the position of spins by virtue of their phase frequencies. The k-
space formalism described below provides an alternate under-
standing of the process of phase encoding.
Using a constant gradient Gx, the precession frequency at loca-
tion x is:
(A.I)
(The relation to equation 2.3 will be noted; equation A.I, how-
ever, refers to the rotating frame, with the result that the static
field component is excluded.)
Phase Encoding 157
image (i.e., to the field of view). Thus along the x and y direc-
tions, FOV"y is
1
FOV"l/=-- (A.6)
- kxy,min
Ciassical2DFT Imaging
We now see how the concept of Fourier space applies to conven-
tional spin-echo imaging. In conventional spin-echo imaging,
spatial encoding is obtained through frequency and phase en-
coding as described in Chapter 2. During the pulse sequence, the
data points are recorded during the readout period, and data
acquisition is repeated for successive values of the phase-encode
gradient. These data points, which are acquired sequentially,
form the k-space image. This section describes the congruence of
the data points to that of the k-space.
At the beginning of the pulse sequence, the spatial frequency is
o in both x and y directions, corresponding to the center of the
k-space (point A in Figure A.3). Typically, a gradient pulse is
applied in the first half of the spin-echo sequence on the so-called
readout axis. Let us call this axis x. The amplitude and duration
of this pulse are fixed and determined, so that the corresponding
kx is kx,maJ2, which is determined by image resolution. Simulta-
neously, a gradient pulse of variable amplitude is applied on the
y axis for phase encoding, corresponding to a spatial frequency
kyo Therefore, after these two pulses we have traversed to posi-
tion B in Figure A.3 (at the extremity of a line in k-space). As long
as no other gradient pulse is applied either on x or y axis, the
spatial frequencies remain the same. The effect of the 180 0 RF
pulse is to invert the phase of the transverse magnetization,
which in k-space results in symmetric reflection with respect to
the center. We move now to position C in k-space. During the
echo, a gradient pulse is applied (readout gradient). Signal inten-
Classical 2DFT Imaging 159
Ky
~.-~~~~~~~~~ C
c'
Kx
1
kx,mm. r xt =
='~
L7
--
FOV (A.8)
x
(A.9)
which means that the area of the gradient pulse applied on the
readout axis in the first period is identical to that of the readout
gradient pulse from its start to the top of the echo. From the top
of the echo, the signal is sampled symmetrically over nJ2
points, up to the end of the kx line (i.e., point E). Hence, the total
duration of the readout gradient pulse is ni. On each side of the
160 Appendix 1 k-Space Formalism
top of the echo, the signal drops according to T2*. One can see
how the gradient pulse intensity, its duration, and the sampiing
rate are tied to the image resolution and the field of view. Shorter
sampling times require stronger gradient pulses.
At this stage of the acquisition process, we have seen that the
sampling of one echo signal for a few milliseconds provides data
for a full line in k-space. To get data on all lines, we must repeat
the whole process again, line after line. To change lines, the
amplitude of the phase-encoding gradient is changed, where-
upon another value of ky is obtained, leading to another point B'
in k-space. Basically, the phase-encoding gradient amplitude is
changed, ensuring that the corresponding spatial frequencies
will vary between -ky,max /2 and +ky,max /2 by ny steps of ky,mix.
Dramatic reduction in scan time can be obtained by changing
the algorithm used to scan the k-space. As seen above, conven-
tional techniques as well as gradient-echo imaging techniques
require the ny lines of the k-space to be scanned individually.
There are, however, other acquisition schemes in which several
lines in the k-space are scanned during a single sequence cycle.
This is the approach used in the pulse sequences based on RARE
(rapid acquisition by refocused echo). Ultimately, the whole k-
space may be scanned in a single cycle, as with echo-planar
imaging (EPI).
Additional Reading
Twieg DB. The k-trajectory formulation of the imaging process with applica-
tions in analysis and synthesis of imaging methods. Med Phys 10:610 (1983).
Index