Pulse Design for Parallel MR Transmission Outline Kspace analysis of small tipangle excitation RF shimming and Parallel MR Transmission B 1 mapping Accelerated MR excitations Homogeneity and SAR minimization ID: 277431
Download Presentation The PPT/PDF document "G16.4427 Practical MRI 1" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
G16.4427 Practical MRI 1
Pulse Design for Parallel MR TransmissionSlide2
Outline
K-space analysis of small tip-angle excitation
RF shimming and Parallel MR Transmission
B
1
mapping
Accelerated MR excitations
Homogeneity and SAR minimizationSlide3
K-Space Interpretation of Small Tip Angle Excitation
New point of view for analyzing selective excitation
Similar approach as that of data acquisition and reconstruction
Strictly valid only for small tip angles excitations but holds also for flip angles ~ 90°
Bloch equation in the rotating frame
(ignoring T
1
and T
2
)Slide4
Excitation K-Space
(parametrically describes a path through the spatial frequency space)
Parametric description of the unit-weight trajectory (
sampling structure)
Spatial-frequency weighting of the
k
-space trajectorySlide5
Application to Slice-Selective Excitation
Conventional slice-selective excitation pulse sequence
k
-space interpretation
K-space is scanned linearly as the RF field is applied
The location in
k
-space at time
t
is the integral of the remaining gradient waveform
Origin reached when the remaining waveform integrates to zero
RF weighting is symmetric with respect to the origin
Slice profile (Fourier transform of RF weighting) is in phase
The role of the refocusing lobe is to shift the
k
-space origin back in the middle of the RF excitation Slide6
Multiple Coil Excitations
a
1,
φ
1
a
2,
φ
2
a
8,
φ
8
a
7,
φ
7
a
6,
φ
6
a
5,
φ5
a3,
φ3
a4,
φ4
RF
RF
RF
RF
RF
RF
RF
RF
RF
a
2
(t)
,
φ
2
(t)
a
3
(t)
,
φ
3
(t)
a
4
(t)
,
φ
4
(t)
a
1
(t)
,
φ
1
(t)
a
7
(t)
,
φ
7
(
t)
a
6
(t)
,
φ
6
(t)
a
5
(t
)
,
φ
5
(t)
a
8
(t)
,
φ
8
(t)
RF shimming
Distinct but time-constant amplitudes and phases for each element
Common gradient and RF waveform
Parallel Transmission
Distinct and time-varying amplitudes and phases for each element
Common gradient waveform but distinct RF waveformSlide7
Parallel RF Transmission
Parallel transmission may be used to correct RF inhomogeneities, control SAR, tailor excitations
Requires calibration of coil array excitation patterns, and operates in close analogy to parallel receptionSlide8
Small Flip Angle Excitation
Homogeneous volume coil
excitation
k
-space sampling trajectory
(controlled by the switching gradients)
spatial-frequency weighting
(proportional to the coil driving current)
Transmit coil array
(
M
xy
is obtained by multiplying the profile by
iγM
0
)
B
1
spatial
weighting
Question:
what is the B
1
spatial weighting?
Effective spatial weighting, to account for coupling-induced
intercoil
correlationsSlide9
Illustration of Parallel Transmission
1 RF pulse
+ the gradient pulseSlide10
Illustration of Parallel Transmission
1 RF pulse
+ the gradient pulse
+ B
1
weightingSlide11
Illustration of Parallel Transmission
L (coils) RF pulses
+ the gradient pulse
+ B
1
weightingSlide12
Example: 2D Selective ExcitationSlide13
Example: 3D Selective ExcitationSlide14
Outline
K-space analysis of small tip-angle excitation
RF shimming and Parallel MR Transmission
B
1
mapping
Accelerated MR excitations
Homogeneity and SAR minimizationSlide15
B1
Mapping
Accurate transmit RF field (B
1
+
) or flip angle maps are needed for many MR applications.
Examples?
Correct the results of quantitative methods Validate theoretical models for EM calculations
Testing MR compatibility of implanted objects Compensate for B1 inhomogeneitiesImage-based RF field measurements are needed for in-vivo
applications
Several B
1
mapping techniques exists, but further improvements (time efficiency, anatomical coverage, accuracy) are needed to use them in the routine practice and for parallel transmissionSlide16
Multi-Point Intensity Method
Non-selective RF pulse (long TR) and FID
Signal is largely independent from T
1
and T
2S ∝sin(α)
Step through transmit voltage until the first signal maximum is found (i.e. α = 90°)
Other pulse amplitudes would then be set relative to this calibration pulseFor GRE:Slide17
Double Angle Method (DAM)
Collect two scans, one of which uses twice the RF amplitude of the other.
image value at
pixel
j
object magnetization
at voxel
j
unknown actual
flip angle
error
“Double angle formula”Slide18
Double Angle Method (DAM)
Collect two scans, one of which uses twice the RF amplitude of the other.
image value at
pixel
j
object magnetization
at voxel
j
unknown actual
flip angle
error
Inefficient method (TR
≥
5T
1
required)
Performs poorly in regions of low signal
ambiguities
if
is
too large, sensitive to noise if
is
too small
“Double angle formula”Slide19
Phase-Based Method
Exploits the fact that rotations do not commute
Final M
xy
differs by a phase
that depends on the magnitude of the
flip angle
α
α
x
α
y
α
-
x
α
-
y
α
x
α
y
α
y
α
x
α
-
y
α
-
x
α
y
α
xSlide20
Question:
What are pros and cons of the phase-based method?Slide21
Phase-Based Method
Exploits the fact that rotations do not commute
Final M
xy
differs by a phase
that depends on the magnitude of the
flip angle
α
Works
better for larger
α
and shorter pulses
(
SAR limitation)
Only for 3D and sensitive to motion/flow
α
x
α
y
α
-
x
α
-
y
α
x
α
y
α
y
α
x
α
-
y
α
-
x
α
y
α
xSlide22
Actual Flip Angle Imaging (AFI)
T
wo identical RF pulses followed by two delays of different
duration (TR
1
< TR
2
< T
1
)
Assumption: at the end of both TR
1
and TR
2
the transverse magnetization is completely spoiled (need RF spoiling with dummy repetition to reach steady state)
Before each excitation pulse:
The observed signals are:
Their ratio is:Slide23
Any questions?Slide24
Outline
K-space analysis of small tip-angle excitation
RF shimming and Parallel MR Transmission
B
1
mapping
Accelerated MR excitations
Homogeneity and SAR minimizationSlide25
Parallel Transmit For 2D EPI Excitation
Complex-valued excitation profile
Periodic excitation pattern associated with the RF pulse of the
l
th
transmit coil
In the case of a 2D EPI excitation trajectory, let’s define:
If the sampling interval in excitation
k
-space is sufficiently small, then
Δx
= 1/Δ
k
x
is big enough that all the aliasing lobes are outside the FOV:
If we
undersample
(i.e. use a larger sampling interval) in excitation
k
-space, then
M
lobes will alias inside the FOV:Slide26
Accelerated Parallel MR Excitations
(in the case of the EPI excitation trajectory we can treat each position separately)
To design our parallel transmit pulse design we need to find the periodic excitation patterns for each transmit coil such that:
Question:
how would the equation above change for an accelerated RF excitation? Slide27
Accelerated Parallel MR Excitations
(in the case of the EPI excitation trajectory we can treat each position separately)
To design our parallel transmit pulse design we need to find the periodic excitation patterns for each transmit coil such that:
We can exploit the extra degrees of freedom to under sample the excitation by a factor
M
: Slide28
Outline
K-space analysis of small tip-angle excitation
RF shimming and Parallel MR Transmission
B
1
mapping
Accelerated MR excitations
Homogeneity and SAR minimizationSlide29
SAR and RF Homogeneity
SAR management and RF homogeneity are critical issues at high magnetic field strengths
SAR is a potentially elevated
safety concern
B
1
focusing compromises
the underlying SNR increaseSlide30
RF Power Deposition
in Multiple Coil Excitations
electric field covariance matrix
RF energy dissipated in
noise covariance matrix
Net Electric Field
EPI excitation trajectory
Small tip angle
Global SAR
Image-domain global SAR
unit current electric field
RF excitation patternsSlide31
Pulse Design for SAR Reduction
weighting
Homogeneous excitation
with minimum SAR
target excitation
profile at
Minimum global SAR
Optimal excitation patterns
for Parallel Transmission
Time-independent RF shimming
Optimal modulation coefficients
shared excitation profile
optimal modulation Slide32
Minimum SAR with Parallel
Tx
SAR =
3.3
20 coils
SAR =
5.5
12 coils
SAR =
7.9
8 coils
Parallel Transmission - 7 Tesla - No Acceleration
SAR =
1
Ultimate
Basis SetSlide33
SAR vs. Profile Homogeneity
B
o
= 7TSlide34
Calibrating the Phi Matrix
2
3
4
1
3
4
1
2Slide35
Calibrating the Phi Matrix
2
4
1
3
3
1
2
4Slide36
Calibrating the Phi Matrix
4
1
3
2
3
4
1
2Slide37
How Many Measurements?
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Experiment
Channel
1
2
3
4
4 real
diagonal elements
6
complex
off-diagonal elements
16 variables to estimate
16 (# of channels × # of channels) measurements required
Question:
Why 16 measurements?Slide38
Simulation Results
1
2
3
4
Φ
calibrated using
simulated E fields
Φ
calibrated using
power measurements
Identical !
Measured and predicted power
Simulation set upSlide39
Experimental Set Up (7 Tesla)
Directional couplers
RF switch
National Instrument Dual 16x1 MUX
Knee setup,
8-channel parallel
Tx
stripline
coil
Power meter
Rhodes & Schwarz
NRP
-Z11
Computer that automates the measurementSlide40
In-Vivo Results
40-miliseconds
Φ
calibrated at 60V
40-miliseconds
Φ
calibrated
at
120V
measured
power
predicted
powerSlide41
Uses of Phi Matrix Calibration
Prediction and real-time monitoring of global SAR
Prediction and real-time monitoring of individual channel FWD and RFL power
Real-time detection of
Tx
chain hardware failures
Optimization of RF pulse design for RF shimming and parallel transmissionSlide42
Maximum Efficiency RF Shimming
Array transmit efficiency metric:
Average B
1
+
strength squared:
Total power deposition
# of spatial locations
It can be treated as a generalized
eigenvalue
problem:
Largest
eigenvalue
= maximum transmit efficiency
η
max
Corresponding eigenvector =
w
max
for maximum efficiency RF shimming
Maximum Efficiency RF Shimming
Find
w
that maximize
ηSlide43
Experiment: Hip Imaging at 7 T
Flip Angle Maps
90
0
π
-π
Excitation: 4
ch
Tx
/Rx loop coils
Receive: 10
ch
Tx
/Rx
(5 loop/
stripline
modules)
Conservative parallel transmit SAR limits were used
Φ
matrix computed from forward and reflected power measures
RF Shim weights
Γ
- matrix
No Shimming
Φ
- matrixSlide44
RF Shimming Weights Calculation
B
1
+
map acquisition and field extraction: ~1 min
Φ
matrix calibration:
< 5 seconds
Maximum
efficiency weights calculation
: < 1
sSlide45
Results: RF Shimming at 7 T
No RF Shimming
Maximum Efficiency RF Shimming
Transmit Efficiency (
η
)
117
286
Measured Total Average Energy Deposition (Watts)
155
58.8
Mean Flip angle in ROI
28.8° ± 10.7°
25.1° ± 10.9°Slide46
Any questions?Slide47
See you after Spring Break!