Resonatoren - Medizinische Fakultät Mannheim

Transcription

Resonatoren - Medizinische Fakultät Mannheim
RUPRECHT-KARLSUNIVERSITY HEIDELBERG
RUPRECHT-KARLSUNIVERSITY HEIDELBERG
Computer Assisted Clinical Medicine
Dr. Friedrich Wetterling
Computer Assisted Clinical Medicine
Dr. Friedrich Wetterling
6/20/2011 | Page 1
Hochschule Mannheim
6/20/2011 | Page 2
MRI Components: Physical Parameters
Bildgebende Systeme in der Medizin
Magnetresonanztomographie IV:
radiofrequency RF
Radiofrequenz Resonatoren
gradients Gxyz
shim coils
static field B0
Dr. Friedrich Wetterling
transmitter
shim
receiver
Computer Assisted Clinical Medicine
Faculty of Medicine Mannheim
University of Heidelberg
Theodor-Kutzer-Ufer 1-3
D-68167 Mannheim, Germany
Friedrich.Wetterling@MedMa.Uni-Heidelberg.de
www.ma.uni-heidelberg.de/inst/cbtm/ckm/
350 MHz
control
panel
computer
RUPRECHT-KARLSUNIVERSITY HEIDELBERG
RUPRECHT-KARLSUNIVERSITY HEIDELBERG
Computer Assisted Clinical Medicine
Dr. Friedrich Wetterling
Computer Assisted Clinical Medicine
Dr. Friedrich Wetterling
6/20/2011 | Page 3
MRT V:
350 MHz
6/20/2011 | Page 4
image
processor
gradient
technical
component
Æ
static field B0
Æ
physical
parameter
M0
radiofreq. RF Æ
signal
gradients Gxyz Æ
image
Maxwellgleichungen 3 + 4
(Dynamische Felder)
3. Ein magnetisches Wechselfeld erzeugt ein elektrisches Feld (Induktionsgesetz)
v
v
∂B
∇ ×E = −
∂t
v v
∂Φ B , S
∫ E ⋅ dl = − ∂t
∂S
Doppelresonanter 1H/23Na Kopfresonator für
Gehirnuntersuchungen.
4. Magnetische Felder werden durch gerichtete Ladungsbewegung erzeugt (Bio-Savart)
1H
Bild vom Kopf.
TIM TRIO mit 23Na Resonatorsystem für MRT bei 3T.
Ganzkörper
Menschen
23Na
v
v v ∂D
∇ × H = Jf +
∂t
v v
∂Φ D,S
∫ H ⋅ d l = If ,S + ∂t
∂S
23Na
Bild vom Kopf mit
eingezeichneter Spule
(schwarz).
Bild des
RUPRECHT-KARLSUNIVERSITY HEIDELBERG
RUPRECHT-KARLSUNIVERSITY HEIDELBERG
Computer Assisted Clinical Medicine
Dr. Friedrich Wetterling
Computer Assisted Clinical Medicine
Dr. Friedrich Wetterling
Resonanzschwingkreis
Impedanz eines Kondensators:
Z L = j ωL
Z C = 1 / j ωC
Resonanzfrequenz eines idealen Resonators:
ω=
S rel = 1V ⋅ 10
−
A ( s21 )
20
Measured relative sensitivity profile coincides well with
measured MRI and theoretically calculated sensitivity
profiles
1
LC
Frequency [MHz]
ω = 2πf
S21 att. [dB]
Impedanz einer Induktivität:
Relative Signal Sensitivity
of a Surface Coil
6/20/2011 | Page 6
S11 att. [dB]
6/20/2011 | Page 5
Frequency [MHz]
Seite 1
1
RUPRECHT-KARLSUNIVERSITY HEIDELBERG
Computer Assisted Clinical Medicine
Dr. Friedrich Wetterling
RUPRECHT-KARLSUNIVERSITY HEIDELBERG
Signal Sensitivity Profile
of a Surface Coil
Computer Assisted Clinical Medicine
Dr. Friedrich Wetterling
6/20/2011 | Page 7
Auskopplung des der induzierten Spannung:
50Ohm-Anpassung (Matching)
6/20/2011 | Page 8
ω =
1
LC
Measured NMR signal
S∝
2
(y
r2
2
+ r2
Capacitive tuning (variable)
•
Capacitive matching (variable)
C t1 / 2 =
L
L
±
R eff2 + ω 2 L2 R eff2 + ω 2 L2
RUPRECHT-KARLSUNIVERSITY HEIDELBERG
Computer Assisted Clinical Medicine
Dr. Friedrich Wetterling
Messung des Qualitätsfaktors
Δω−3dB (s11 )
⎛
1 ⎞
2
⎟
R eff + ⎜⎜ ω L −
ω C t ⎟⎠
⎝
L
ω 2 L2 −
+ R eff2
Ct
L2 −
(R
i
(
− R eff )⋅ R eff2 + ω 2 L2
)
ω 2 Ri
2
Ct <
L
ω 2 L2 + R eff 2
3
RUPRECHT-KARLSUNIVERSITY HEIDELBERG
6/20/2011 | Page 10
Q = 2⋅
…coil capacitance
)
Computer Assisted Clinical Medicine
Dr. Friedrich Wetterling
6/20/2011 | Page 9
…larmor frequency
… coil inductance
C
•
C m = −Ct
depends on the coil geometry
ωL
L
Empirische Detektoroptimierung bei 79MHz
ω0
Δω−3dB (s11 )
-je breiter der
ω0
Resonanzpeak, desto
-kleinere Detektoren haben
schlechter die
höhere Sensitivität in
Sensitivität (bei gleicher
Detektorzentrum
Spulengeometrie)
s11-Reflexionsmessung mit einem Netzwerkanalysator am Port 1.
-größere Detektoren haben
höhere Sensitivität in
tieferliegenden Regionen
Wetterling, PhD Thesis, Trinity College Dublin, 2009
RUPRECHT-KARLSUNIVERSITY HEIDELBERG
Computer Assisted Clinical Medicine
Dr. Friedrich Wetterling
Wetterling, PhD Thesis, Trinity College Dublin, 2009
Beispiel I:
Double-Tuned 23Na/ 1H Surface Coil
RUPRECHT-KARLSUNIVERSITY HEIDELBERG
Computer Assisted Clinical Medicine
Dr. Friedrich Wetterling
6/20/2011 | Page 11
Beispiel I:
23Na/ 1H Surface Coil MRI Results
6/20/2011 | Page 12
•
100% SNR gain in 10mm sample depth
(due to anatomical coil shape and Q-factor improvement)
•
23Na
Commercial TXRX
Developed TXRX
Surface coil
Surface coil
channel:
•
- 50 % SNR gain at the surface
In vivo 1H MRI:
5min acquisition time, 200µm-plane
(due to Q-factor improvement )
resolution and 2mm slice thickness
- add. 30 % SNR gain in 10mm depth
(due to anatomical coil shape)
•
1H
channel:
•
- 0 % SNR gain at the surface
In vivo 23Na MRI:
40min acquisition time, 500µmin-plane
(due to Q-factor improvement )
resolution and 2mm slice thickness
- 20 % SNR gain in 10mm depth
(after two-fold zero filling)
(due to anatomical coil shape)
Seite 2
2
RUPRECHT-KARLSUNIVERSITY HEIDELBERG
Computer Assisted Clinical Medicine
Dr. Friedrich Wetterling
RUPRECHT-KARLSUNIVERSITY HEIDELBERG
Beispiel II: Double-Tuned 23Na/ 1H
Volume Resonator
Computer Assisted Clinical Medicine
Dr. Friedrich Wetterling
6/20/2011 | Page 13
•23Na/1H
Beispiel II: Double-tuned 23Na/ 1H
Volume Resonator/ MRI Results
6/20/2011 | Page 14
12-leg birdcage structure with RF-shielding
•
200% SNR loss in 10mm depth
•Optimized for transmit homogeneity ( ± 7 % ; required for qNa- MRI)
BUT ± 7 % signal homogeneity across the entire sample
•Variable balanced tuning/matching via two mechanically coupled trimmer capacitors
Commercial TXRX
Developed TXRX
surface coil
volume coil
•Active decoupling
•
In vivo 1H MRI:
5min acquisition time, 300µm-plane
resolution and 2mm slice thickness
RUPRECHT-KARLSUNIVERSITY HEIDELBERG
Computer Assisted Clinical Medicine
Dr. Friedrich Wetterling
RUPRECHT-KARLSUNIVERSITY HEIDELBERG
Beispiel III:
Receive-Only 23Na Surface Resonator
Beispiel III:
23Na Surface Resonator/ MRI Results
Computer Assisted Clinical Medicine
Dr. Friedrich Wetterling
6/20/2011 | Page 15
6/20/2011 | Page 16
Commercial TXRX
Developed TORO
surface coil
volume/ surface coil
•
0% SNR gain in 10mm depth
•
Homogeneous sample excitation (doubled penetration
depth)
•
• Receive-only surface coil
In vivo 23Na MRI:
40min acquisition time, 500µm in-plane
• Optimised for SNR
resolution and 2 mm slice thickness
• Active decoupling
(after two-fold zero-filling)
• 200 % SNR gain in 10 mm sample depth
RUPRECHT-KARLSUNIVERSITY HEIDELBERG
RUPRECHT-KARLSUNIVERSITY HEIDELBERG
Computer Assisted Clinical Medicine
Dr. Friedrich Wetterling
Computer Assisted Clinical Medicine
Dr. Friedrich Wetterling
6/20/2011 | Page 17
Coil System Comparison
Beispiel IV:
Vorverstärkerentkoppelte Detektoren
6/20/2011 | Page 18
Acquisition Parameter:
•
Preamplifier Matching
•
Active Decoupling
2D FLASH,
Tacq = 10min, TE = 3ms, TR = 250ms
25mm x 25mm
square loop, 2mm wide,
0.5mm thick silver foil
(fR=79MHz)
1.25mm in plane res., 1mm ST,
non-physiological 4M NaCl solution
79 MHz
10mm
150
50
Receive-Only (RO)
Surface Coil
Transceiver (TXRX)
Surface Coil
Transceiver (TXRX)
Birdcage Coil
SNR in au
40 dB
100
Active Decoupling
Loop (fR=79MHz)
[A. Reykowski et al,
Magnetic Resonance in
Medicine, 1995. 33(6): p.
848-852]
Preamp Matching
Loop (fR=79MHz)
Low Input
Impedance
Preamplifier
Input circuit
of preamplifier
(+) limited current flow in the coil Æ no interaction with adjacent coils
(+) limited dielectric losses in object: Non-resonant Æ no-tune no-match
(-) complex circuit diagram
(-) receive-only (!) Æ separate transmit resonator necessary
0
Seite 3
3
RUPRECHT-KARLSUNIVERSITY HEIDELBERG
RUPRECHT-KARLSUNIVERSITY HEIDELBERG
Computer Assisted Clinical Medicine
Dr. Friedrich Wetterling
Computer Assisted Clinical Medicine
Dr. Friedrich Wetterling
6/20/2011 | Page 19
Beispiel V: Multi-element Detektoren
Multi-element Detektoren: Summation
mittels Sum-of-Squares Methode
6/20/2011 | Page 20
2
2
(
+
SQRT
)=
Figure 1: The schematic circuit diagram for one
element of the two-element phased array coil with
both preamplifier and active decoupling.
Figure 2:
Newly-developed 23Na receive-only twoelement phased array coil in (left) bottom, and (right) side
view.
RUPRECHT-KARLSUNIVERSITY HEIDELBERG
RUPRECHT-KARLSUNIVERSITY HEIDELBERG
Computer Assisted Clinical Medicine
Dr. Friedrich Wetterling
Computer Assisted Clinical Medicine
Dr. Friedrich Wetterling
6/20/2011 | Page 21
Parallel Imaging: Coil Sensitivity
6/20/2011 | Page 22
Parallel Imaging: Receive Coils
102 seamlessly integrated coil elements
at 32 receiving channels
matrix coils:
head
• surface coils
• inhomogeneity correction
• phased array coils
neck
• image combination
• parallel imaging: SMASH / SENSE
stem
leg
array body
combination
coil
courtesy: Siemens AG, Erlangen
RUPRECHT-KARLSUNIVERSITY HEIDELBERG
RUPRECHT-KARLSUNIVERSITY HEIDELBERG
Computer Assisted Clinical Medicine
Dr. Friedrich Wetterling
Computer Assisted Clinical Medicine
Dr. Friedrich Wetterling
6/20/2011 | Page 23
Parallel Imaging: Receive SENSE
6/20/2011 | Page 24
Parallel Imaging: Receive SMASH I
k-space
c
constant
cos Δ k y y
y
coil 1
coil 3
coil 4
sin Δ k y y
cos 2 Δ k y y
- SENSE (sensitive encoding) works in the image domain
- use sensitivity and location information of coils to unfold image
- 32 receiving channels standard at modern MRI scanners
Pruessmann et al. MRM 1999
coil 2
128 channel prototype, MGH Boston
- SMASH (simultaneous acquisition of spatial harmonics)
works in the k-space
- use sensitivity and location information of coils
to reconstruct missing lines (harmonics) of the k-space
- 32 receiving channels standard at modern MRI scanners
sin 2 Δ k y y
Sodickson and Manning. MRM 1997
Seite 4
4
RUPRECHT-KARLSUNIVERSITY HEIDELBERG
RUPRECHT-KARLSUNIVERSITY HEIDELBERG
Computer Assisted Clinical Medicine
Dr. Friedrich Wetterling
Computer Assisted Clinical Medicine
Dr. Friedrich Wetterling
6/20/2011 | Page 25
coil 1
Parallel Imaging: Receive SMASH II
coil 2
6/20/2011 | Page 26
Parallel Imaging: Receive SMASH III
coil 3
- speed
cMRA: 18 s
1. harmonic
0. harmonic
- resolution
0.4 x 0.4 x 1 mm3
courtesy: Siemens AG, Erlangen
Sodickson and Manning. MRM 1997
RUPRECHT-KARLSUNIVERSITY HEIDELBERG
RUPRECHT-KARLSUNIVERSITY HEIDELBERG
Computer Assisted Clinical Medicine
Dr. Friedrich Wetterling
Computer Assisted Clinical Medicine
Dr. Friedrich Wetterling
6/20/2011 | Page 27
Whole Body MRI
6/20/2011 | Page 28
Whole Body MRI: Metastasis Screening
areas of applications:
Oncology D
Cardiovascular D
Inflammatory/ Rheumatism ~
Infection/ Immunology
Health Care Screening
• complex, multiparametric classification with local
and global parameters
• training systems
• computer aided diagnosis vs. automated diagnosis
RUPRECHT-KARLSUNIVERSITY HEIDELBERG
RUPRECHT-KARLSUNIVERSITY HEIDELBERG
Computer Assisted Clinical Medicine
Dr. Friedrich Wetterling
Computer Assisted Clinical Medicine
Dr. Friedrich Wetterling
6/20/2011 | Page 29
Whole Body Moving Table Angiography
6/20/2011 | Page 30
23Na-birdcage
Whole-body 23Na MRI (12 000 times less
signal compared to 1H-MRI)
on patient bed
(outside 3T magnet)
Split 23Na-birdcage
courtesy: Siemens AG, Erlangen
Wetterling et al., ISMRM 2011
Seite 5
5
RUPRECHT-KARLSUNIVERSITY HEIDELBERG
RUPRECHT-KARLSUNIVERSITY HEIDELBERG
Computer Assisted Clinical Medicine
Dr. Friedrich Wetterling
Computer Assisted Clinical Medicine
Dr. Friedrich Wetterling
6/20/2011 | Page 31
Radiofrequenz-Verluste
6/20/2011 | Page 32
Detektorkühlung lohnt sich, wenn ohmsche
Verluste überwiegen
-Q-Faktoren der parasitischen Kapazität dominieren
Verluste, wenn die Impedanz einen Stromfluß durch das
-Überwiegen die ohmschen Verluste in der Spule, kann ein signifikanter
Objekt erlaubt (Z klein, Abhängig von Frequenz,
Sensitivitätsanstiege durch Kühlung der Spule erzielt werden (Kühlung lebender
Geometrie und Leitfähigkeit des Objekts)
Subjekts bei MRT ist unmöglich).
- Test, ob sich Kühlung lohnt: Q-Faktormessung mit gewählter Spulengeometrie
und Beladung
Qunbeladen ∝
ZC =
1
jωC
Qbeladen
L 1
C ROhm
Rauschen = 4 ⋅ ( ROhm + R parasitisch ) ⋅ k B ⋅ T ⋅ Δf
L
1
∝
C ROhm + R parasitisch
A( Fläche )
C∝
d ( Durchmesser )
Wetterling, PhD Thesis, Trinity College Dublin, 2009
RUPRECHT-KARLSUNIVERSITY HEIDELBERG
Computer Assisted Clinical Medicine
Dr. Friedrich Wetterling
6/20/2011 | Page 33
Spulenvergleich bei 100MHz und ~15mm
Durchmesser
RUPRECHT-KARLSUNIVERSITY HEIDELBERG
Computer Assisted Clinical Medicine
Dr. Friedrich Wetterling
6/20/2011 | Page 34
High Field MRI: RF Problems
wavelength at 130 MHz in air: 2.3 m
RF
wavelength in body: ~ 0.25 m
SNR
c) Cryo:
92
b) Eigenbau:
14.9
a) Commercial:
7.9
RF
problem: dielectric resonances
Sack, Wetterling et al., ISMRM 2011
RUPRECHT-KARLSUNIVERSITY HEIDELBERG
RUPRECHT-KARLSUNIVERSITY HEIDELBERG
Computer Assisted Clinical Medicine
Dr. Friedrich Wetterling
Computer Assisted Clinical Medicine
Dr. Friedrich Wetterling
6/20/2011 | Page 35
High Field MRI: RF Problems 3 Tesla
6/20/2011 | Page 36
Zusammenfassung
1. Abstimmen und Anpassen von Resonanzschwingkreis für MRT
2. Charakterisierung von MRT Resonatoren
darkening can
be avoided by
dielectric pad
3. Optimierung der Detektorsensitivitäten
4. Oberflächenspulen (Leistungsangepass/ Rauschangepasst)
5. Volumenresonator
6. Transmit-Only Receive-Only (TORO) Systeme
- wavelength has the same dimension as organs in the body
→ reflections at organ borders create interference called “dielectric resonance”
7. Multi-element (Phased-Array) Detektorsysteme (Ganzkörperbildgebung)
- RF absorption increases
8. Vorteil durch Detektorkühlung
- B1 – field is inhomogeneous
9. Probleme bei hohen Feldstärken Æ Sende-Arrays notwendig
- dielectric resonances appear at transmitting and receiving
courtesy: Hennig. Department of Diagnostic Radiology, University Freiburg
Seite 6
6