magnetic resonance imaging of fast relaxing spins: acquisition during adiabatic excitation november...
TRANSCRIPT
Magnetic Resonance Imaging of Fast Relaxing Spins:
Acquisition during Adiabatic Excitation
November 14, 2005, CMRR : Djaudat Idiyatullin
Mike’s crazy idea is working
Interleaved excitation and sampling during a frequency-swept pulse
1. Steady state 2. Sensitive to spins with a very
short T23. It is not FID but the signal
predictable by Bloch simulation.
d
p
. . .
BIR4
……
How to extract information from this weird sampling during swept excitation?
Least square method Monte Carlo simulation Wavelet transform
22 1aT
21/ 1a
How to extract information from this weird sampling during swept excitation?
Solution:
1. Away from adiabatic condition
22 1aT
21/ 1a
2 21 1a a
2. Linear system- Correlation method
0 0180 90
Linear system
Output r(t)
Input x(t)
System h(t)
A system is linear if:1. Linearity : Output = C * Input2. Shift invariant : delaying of Input → same delaying of Output
0 τ t
x(τ) → x(τ)h(t- τ)( ) ( ) ( )r t x h t d
( ) ( ) ( )R X H
Convolution
Fourier theorem:
Evolution of the isochromats during HS8 pulse (dw=10mks, (~ 30 degree), R2=500Hz)
30kHz
time
20kHz
10kHz
0kHz
-10kHz
-20kHz
-40-20
02040
30kHz
frequency
20kHz
10kHz
0kHz
-10kHz
-20kHz
sum of all
Evolution of the isochromats during HS8 pulse (dw=10mks, (~ 30 degree), R2=500Hz)
30kHz
time
20kHz
10kHz
0kHz
-10kHz
-20kHz
-40-20
02040
30kHz
frequency
20kHz
10kHz
0kHz
-10kHz
-20kHz
sum of all
Linear system
Correlation method for linear system
( ) ( ) ( )r t x h t d
Response r(t)
Excitation x(t)
Spin system h(t) FT
FT
( ) ( ) ( )R X H *
2
( ) ( ) ( )( )
( )( )
R X RH
XX
( )H
)(X
( )R
System spectrum
*
Simulated data HS4 pulse
100 isochromats from -12.5kHz step 250Hz
dw=10mks R1=500Hz
-40 -20 0 20 40
Hy()
f
/2 , kHz
eH
x()
dX
x()
cR
x()
bR
xy()
a
spin density
SWeep Imaging
with Fourier Transform (SWIFT)
(a)
Gz
Gy
Gx
acq
RF
1
RF
1
acq
(b)
. . .
Tp
Tr
HSn pulsesFlip angle < 90 degreeTr ~ TpBw=sw=2πN/TpBack-projection reconstruction
SWIFT, characteristics
Signal intensity depends only on T1 and spin density (M0) :
Maximum signal intensity Ernst angle:
Maximum T1 contrast:
Spin density contrast:
Sensitive to short T2 :
10
1
1 exp( / )sin( )
1 exp( / )cos( )tr
tr
T TS M
T T
1cos( ) exp( / )opt trT T
1.7 opt
opt
2 1/ ~ 10T s s
SWIFT, hardware problems
“Dead time” after pulse
4.7T , 7T : ~ 3μs : sw < 130kHz
4T : ~ 20μs : sw < 40kHz
FIFO underflow happens if:
Tr < 5ms for 128 samplingTr < 10ms for 256 samplingsw ~ 25-35 kHz
MIP of 3D image sw=32kHz
128x128 x 644T
Empty “16”-element TEM head coil
3D image of thermoplastic
T2~0.3ms sw=100kHz
128x128 x 1284.7T
Sensitivity to short T2
MIP of 3D image plastic toy in breast coil sw=39kHz
128x128 x 128D=25cm
4T
Sensitivity to short T2
Slices of 3D image of feet
sw=20kHz4T
First in vivo SWIFT 3D images
Slices of 3D image
raspberryin vivo
sw=100kHz128x128x128
D=3cm4.7T
Sensitivity to raspberry
Advantages Disadvantages
fast Too fast for VARIAN
FIFO underflow
Sensitive to short T2 Sensitive to coil material
Reduced motion artifacts(zero echo time, back projection reconstruction)
Problems with slice selection
Reduced signal dynamic range ?
quiet Too quiet
AnotherMike’s
crazy idea
Breast MR scanner
Thanks to:Ivan Tkac Gregor Adriany Peter Andersen Tommy Vaughan Xiaoliang ZhangCarl SnyderBrian Hanna John StruppJanis Zeltins Patrick BolanLance DelaBarreUte Goerke
all CMMR
Fast & Quiet MRI by Sweeping Radiofrequency
Djaudat Idiyatullin, Curt Corum, Jang-Yeon Park, Michael Garwood
Macros, C programming
Hardware
Software
Yellow pages of CMRR Discussion