Вы находитесь на странице: 1из 23

Magnetic Resonance Imaging (MRI) - Chapter 15 25 January, 8 and 15 February, and 1 March 2007

Take Aways: Five Things You should be able


to Explain after the MRI Lectures

How the MR signal is localized within the patient (2D)


Magnetic Resonance Imaging Chapter 15 How the multiple FID echoes are collected ( (k-space
space
data acquisition) and how these are reconstructed into
the grayscale image data visualized on PACS (2D)
How 3D volume data is acquired and reconstructed
Brent K. Stewart, PhD, DABMP What factors of the MRI data collection process play into
Professor, Radiology and Medical Education
the resulting quality of reconstructed image slices and
volumes
Director, Diagnostic Physics
How consideration of artifacts, safety/bioeffects and
instrumentation play into the decisions you will be
a copy of this lecture may be found at:
making in the future with regards to image interpretation,
http://courses.washington.edu/radxphys/PhysicsCourse
http://courses.washington.edu/radxphys/PhysicsCourse.html
.html
magnet operation and system purchase

UW and Brent K. Stewart PhD, DABMP 1 2


UW and Brent K. Stewart PhD, DABMP

Localization of the MR Signal Magnetic Field Gradients (1): how are they made?

Spatial localization requires the imposition of known and Linear magnetic field gradients
controlled magnetic nonuniformities with prescribed directionality
and strength are produced in
Linear gradients are superimposed on the homogeneous paired wire coil configurations
and much stronger main magnetic field (B0) energized with a DC current of
specific polarity and amplitude
The change in Larmor frequency (f0) of the precessing
Gradient null point; reverse
nuclei are used to distinguish position of the NMR signal
grad. polarity w/ opp. current
within the object
Linear over a predefined field
Conventional MRI involves RF excitations (NMR) of view (FOV) usu 50 cm
combined with magnetic field gradients to localize the Three sets: x, y and z; can also
signal from volume elements (voxels
(voxels)) within the patient generate oblique gradient w/
superposition

c.f. Bushberg, et al. The Essential Physics


3 4
UW and Brent K. Stewart PhD, DABMP of Medical Imaging, 2nd ed., p. 416-
416-7. UW and Brent K. Stewart PhD, DABMP

UW and Brent K Stewart, PhD, DABMP 1


Magnetic Resonance Imaging (MRI) - Chapter 15 25 January, 8 and 15 February, and 1 March 2007

Magnetic Field Gradients (2): relating f to z Magnetic Field Gradients (3): f across pixel

gradient: e.g., Gz = B/
Larmor freq. changes along gradient: B/z, Gx = B/
B/x Gradient amplitude and number of samples over the FOV
Location of nuclei along gradient is determined by their frequency
frequency determines the frequency bandwidth across each pixel
[fz = (/2
/2)(B/
B/z)
z)z = (/2
/2)Gzz] and phase ( (z = 2fzt) 10 mT/m (42.58 MHz/T 1T/1,000 mT 1 m/100 cm) = 4258 Hz/cm
Peak amplitude of gradient (G (Gz) field (
(steepness
steepness): [1,80] mT/m Localization of nuclei in 2D requires the application of three distinct
Slew rate ((quickness
quickness of gradient ramping): [5,200] mT/m/msec and orthogonal gradients during the pulse sequence:
sequence: (1) slice select,
select,
(2) frequency encode and (3) phase encode gradients
* From the above
calculation, itits easy to
see that with gradients
friend: /2
our old friend: =
/2
Hz-cm-1/mT-
426 Hz- /mT-m-1,
so then it
its just a
matter of multiplying
the number of mT/m
by this factor to get the
bandwidth (Hz)/cm.
c.f. Bushberg, et al. The Essential Physics c.f. Bushberg, et al. The Essential Physics
5 6
of Medical Imaging, 2nd ed., p. 417. UW and Brent K. Stewart PhD, DABMP c.f. Hashemi, et al.. MRI the Basics, p. 105. of Medical Imaging, 2nd ed., p. 418. UW and Brent K. Stewart PhD, DABMP

Slice Select Gradient SSG (1) Slice Select Gradient SSG (2)

RF pulse antennas can


cant spatially direct the RF energy within FOV For a given gradient strength,
In conjunction with a selective frequency narrowband RF pulse ST determined by RF BW
applied to the entire volume, the SSG determines the imaging sliceslice For fixed RF BW, the gradient
Slice thickness (ST) determined by: ST f/(Gf/(G/2)
/2) strength determines ST
Applied RF pulse bandwidth (BW): f Excite a rectangular slab
Gradient strength across the FOV: G (slice) of nuclei
nuclei sinc
sinc
waveform:
waveform: sinc(t) = sin(t)/t
Need an infinitely long sinc
pulse to get a perfectly
rectangular slice
Truncation in time of applied
RF sinc pulse leads to rounded
and jagged slice profiles

c.f. Bushberg, et al. The Essential Physics c.f. Bushberg, et al. The Essential Physics
7 8
of Medical Imaging, 2nd ed., p. 418. UW and Brent K. Stewart PhD, DABMP of Medical Imaging, 2nd ed., p. 419-
419-20. UW and Brent K. Stewart PhD, DABMP

UW and Brent K Stewart, PhD, DABMP 2


Magnetic Resonance Imaging (MRI) - Chapter 15 25 January, 8 and 15 February, and 1 March 2007

Slice Select Gradient SSG (3) Frequency Encode Gradient FEG (1)

FEG aka: readout gradient


Width of sinc pulse determines Applied to SSG
the RF output frequency BW
Applied throughout formation
Both narrow BW w/ weak and decay of the FID echo
gradient and wide BW w/
from slab excited by the SSG
strong gradient same ST
SNR 1/
1/(BW) fx = (
(/2
/2)Gxx fx x
Narrow BW SNR Demodulation of the composite
Narrow BW chemical shift signal produces a net
frequency variation that is
Gradients cause spin de-
de-
phasing: phase very important! symmetrically distributed from
+fmax to fmax at FOV edges
Re-
Re-establish original phase
with opp. polarity gradient with Spatial projection: column sum
integrated area ( Gt)

c.f. Bushberg, et al. The Essential Physics c.f. Bushberg, et al. The Essential Physics
9 10
of Medical Imaging, 2nd ed., p. 421. UW and Brent K. Stewart PhD, DABMP of Medical Imaging, 2nd ed., p. 422. UW and Brent K. Stewart PhD, DABMP

Frequency Encode Gradient FEG (2) Phase Encode Gradient PEG (1)

Composite signal is amplified, Short duration gradient applied


digitized and decoded by before FEG and after SSG to
Fourier Transform (FT) provide 3rd spatial dimension
After SSG all spins in coherence
f : t, Fourier transform pairs Hz

(like x and spatial frequency) During PEG application linear


variation in precessional
f x FT{FID(t)}
FT{FID(t)} = f x Hz
frequency introducing a persistent
Rotation of FEG direction phase shift across the slice slab in
provides projections through (y Byyt)
PEG direction: (
Hz
object as a function of angle After all FID data collected, a FT is
Like CT: filtered backprojection applied to decode the spatial
()
position along the PE direction (
However, due to sensitivity to
Motion during data collection
motion artifacts phase produces ghosting in along PE
FID(t)
encoding gradients used
Hz

c.f. Bushberg, et al. The Essential Physics c.f. Bushberg, et al. The Essential Physics
11 12
of Medical Imaging, 2nd ed., p. 423. UW and Brent K. Stewart PhD, DABMP of Medical Imaging, 2nd ed., p. 424. UW and Brent K. Stewart PhD, DABMP c.f. Hashemi,
Hashemi, et al.. MRI the Basics, p. 105.

UW and Brent K Stewart, PhD, DABMP 3


Magnetic Resonance Imaging (MRI) - Chapter 15 25 January, 8 and 15 February, and 1 March 2007

Gradient Sequencing Raphex 2001 Diagnostic Questions

For the SE pulse sequence


D43. In MRI, the RF frequency is dependent on the:
Timing of the gradients in conjunction with RF excitation pulses and
data acquisition during echo evolution and decay
Sequence repeated periodically (TR) with only slight changes in the A. Diameter of the body part being imaged
PEG amplitude to provide the 3D identity of protons of the object in
the resulting image
B. Magnetic field strength
C. Pulse sequence
D. Relaxation time
E. RF coil

c.f. Bushberg, et al. The Essential Physics


13 14
of Medical Imaging, 2nd ed., p. 425. UW and Brent K. Stewart PhD, DABMP UW and Brent K. Stewart PhD, DABMP

Raphex 2001 Diagnostic Questions Raphex 2000 Diagnostic Questions

D46. Gradient fields in MRI are principally used to: D48. In MRI images, motion during the scans results in
D48.
ghost images which appear in the ______ direction.
A. Eliminate perturbations in the magnetic field due to
site location A. Amplitude
B. Maintain a uniform magnetic field in the field of view B. Frequency encoding
C. Measure the spin coupling C. Phase encoding
D. Provide spatial localization D. Relaxation
E. Shorten T1 to reduce scan time E. Slice thickness

15 16
UW and Brent K. Stewart PhD, DABMP UW and Brent K. Stewart PhD, DABMP

UW and Brent K Stewart, PhD, DABMP 4


Magnetic Resonance Imaging (MRI) - Chapter 15 25 January, 8 and 15 February, and 1 March 2007

K-space
space Data Acq. and Image Reconstruction Two-
Two-dimensional Data Acquisition
Max. signal in center of k-
k-space
+k +k

MRI data initially stored in a k-


+k
space
space matrix (spatial
frequency domain corr. time :
2D FT

domain; x : k, f : t FT pairs;

Larmor relation through
-k
-k

gradients: k = ( /2)Gxt)
(/2 -k -k -k -k

FID data encoded in kx by FEG


and in ky by PEG With methodical variations of the PEG during each excitation, the the k-
k-
The digitally sampled FID is space matrix is filled (or partially filled) with FID echos
stored along kx in a row MR data acquired as a complex, composite frequency waveform:
-k
corresponding to the -k +k FID(kx,ky) V(t) = V1cos(2ft) + iV2sin(2ft) = Re. + iIm.
magnitude and sign of the
k-space divided into 4 quadrants w/ origin at center
(ky)
PEG (k
Complex conjugate symmetry: only matrix + one line req.

adapted from Bushberg, et al. The Essential adapted from. Hashemi, et al. adapted from Bushberg, et al. The Essential Physics adapted from. Hashemi, et al.
17 18
Physics of Medical Imaging, 2nd ed., p. 426. UW and Brent K. Stewart PhD, DABMP MRI the Basics, p. 140. of Medical Imaging, 2nd ed., pp. 426, 429. UW and Brent K. Stewart PhD, DABMP MRI the Basics, p. 140.

Pulse Sequences Amendment to Bushberg Figure 15-


15-15

Tailoring pulse sequences


emphasizes the image contrast
dependent on , T1 and T2
Timing, order, polarity, pulse
shaping, and repetition
frequency of RF pulses and x,
y and z gradient application
Major pulse sequences
Spin Echo (SE)
Inversion recovery (IR)
Fast Spin Echo (FSE)
Gradient Recalled Echo
(GRE)
Echo Planar Image (EPI)

c.f. http://www.indianembassy.org/dydemo/page3.htm
c.f. Bushberg, et al. The Essential Physics
19 20
UW and Brent K. Stewart PhD, DABMP of Medical Imaging, 2nd ed., p. 428. UW and Brent K. Stewart PhD, DABMP

UW and Brent K Stewart, PhD, DABMP 5


Magnetic Resonance Imaging (MRI) - Chapter 15 25 January, 8 and 15 February, and 1 March 2007

Summary of 2D SE Acquisition Steps (1) Summary of 2D SE Acquisition Steps (2)

(1) Narrowband RF pulse applied (4) Re-


Re-establishment of phase
simultaneously with SSG (center coherence at t = TE (FID echo)
t=0); SSG: z f/(Gz/2)
/2)
(4) During echo formation and
(1) Mz converted to Mxy, the extent
subsequent delay, FEG ( (fx =
determined by the flip
/2Gxx) applied to both
/2
(2) PEG applied to SSG for short
SSG and PEG, encoding
time (encoding precessional
along PE grad.) and with differing
precessional frequency along
amplitudes for each repetition to the readout gradient
create y ( Byyt) along PE (5) Simultaneous to application
direction: multiple views along ky of FEG and echo formation,
(3) Refocusing 180
180 RF pulse the computer acquires the
delivered at t = TE/2: inverting time-
time-domain signal (FID echo)
spins using ADC

c.f. Bushberg, et al. The Essential Physics c.f. Bushberg, et al. The Essential Physics
21 22
of Medical Imaging, 2nd ed., p. 428. UW and Brent K. Stewart PhD, DABMP of Medical Imaging, 2nd ed., p. 428. UW and Brent K. Stewart PhD, DABMP

Summary of 2D SE Acquisition Steps (3) Summary of 2D SE Acquisition Steps (4)

(5) ADC sampling rate determined (8) Object spatial and contrast
by the excitation BW characteristics manifested in
(6) Data stored in k-
k-matrix row (k
(kx) the resulting image
the position (k
(ky) determined by the (8) Final image a spatial
PEG magnitude
representation of the , T1, T2
(6) Inc. changes in PEG mag. fills
and flow characteristics of the
matrix one row at a time (may be
non-
tissues in each voxel using a
non-sequential)
gray-
gray-scale range
(6) When filled partially then copy
complex conjugate data into Voxel thickness determined by
remaining blank rows SSG and RF freq. bandwidth
(7) 2D FT decodes time (spatial Pixel dimension determined by
frequency - k) domain data varying PEG magnitudes and
piecewise along the rows (kx) and readout digitization rate
then columns (ky)

c.f. Bushberg, et al. The Essential Physics c.f. Bushberg, et al. The Essential Physics
23 24
of Medical Imaging, 2nd ed., p. 428. UW and Brent K. Stewart PhD, DABMP of Medical Imaging, 2nd ed., p. 428. UW and Brent K. Stewart PhD, DABMP

UW and Brent K Stewart, PhD, DABMP 6


Magnetic Resonance Imaging (MRI) - Chapter 15 25 January, 8 and 15 February, and 1 March 2007

K-space: the Final Frontier Two-


Two-dimensional Multi-
Multi-planar Acquisition

Bulk of information representing lower spatial frequencies near


center of k-
k-space provides large area contrast in the image Axial (SSG: z, PEG: y, FEG: x)
Higher spatial frequency nearer the periphery provides resolution Coronal (SSG: y, PEG: x, FEG: z)
and detail in the image Sagittal (SSG: x, PEG: y, FEG: z)
Max. signal in center of k--spaceMax. signal in center of k space
Oblique (SSG: a1x + a2y + a3z, etc.)
+k
Data acquisition into the k-
k-space matrix same for all


y z z



-k
x x y
-k +k

c.f. Bushberg, et al. The Essential Physics adapted from Hashemi, et al. c.f. Bushberg, et al. The Essential Physics
MRI the Basics, p. 140.
25 26
of Medical Imaging, 2nd ed., p. 429. UW and Brent K. Stewart PhD, DABMP of Medical Imaging, 2nd ed., p. 430. UW and Brent K. Stewart PhD, DABMP

Acq. Time, 2DFT SE and Multislice Acq. Data Synthesis

Acq. time = TR no. PE steps Take advantage of symmetry


NEX (number of excitations) and redundant characteristics
Example (256x192 matrix, of k-
k-space domain signals
TR=600, NEX=2) 230 sec In PE direction Fourier
Fourier,
PE along lesser matrix NEX
NEX or phase conjugate
dimension to speed acquisition symmetry
symmetry techniques reduce
Multiple slice acquisition also data collection to ky matrix
dimension + 1 line With quadrature detection, have real
speeds image collection and imaginary (90 out of phase)
Max number slices = In FE direction fractional echo
echo components of induced voltage from
TR/(TE+c and read conjugate symmetry
symmetry FID (t):
TR/(TE+c)) V(t) = V1cos(2ft) + iV2sin(2ft)
shorten FID echo sampling
c dependent on MRI system time Two data values per digitized FID
capabilities sample
Both SNR and artifacts Complex conjugate =
Longer TR more slices V1cos(2ft) iV2sin(2ft)

c.f. Bushberg, et al. The Essential Physics c.f. Bushberg, et al. The Essential Physics
27 28
of Medical Imaging, 2nd ed., p. 431. UW and Brent K. Stewart PhD, DABMP of Medical Imaging, 2nd ed., p. 432. UW and Brent K. Stewart PhD, DABMP

UW and Brent K Stewart, PhD, DABMP 7


Magnetic Resonance Imaging (MRI) - Chapter 15 25 January, 8 and 15 February, and 1 March 2007

Inversion Recovery (IR) Acquisition Fast Spin Echo (FSE) Acquisition

180-
180-(TI)-
(TI)-90-
90-(TE/2)-
(TE/2)-180-
180-(TR) FSE uses multiple PE steps w/
SSG, PEG and FEG as SE multiple 180
180 pulses per TR
First echos placed near ky=0
TR long many slices per TR
Best SNR least T2 decay
STIR
Immunity from B0 inhomogen.
Short Tau IR with up to 16x faster collection
Eliminate Fat Lower SNR for high-
high-freq ky
TI = 180 msec Fewer slices collected per TR
FLAIR SE: 8.5 min (TR=2000, 256
FLuid
FLuid Attenuated IR PE)
Eliminate CSF FSE: 2.1 min (TR=2000, 256
TI = 2,400 msec PE steps and 4 echos per TR)
aka: turbo SE
SE & RARE (R (Rapid
Acq. w/ Refocused Echoes)

c.f. Bushberg, et al. The Essential Physics c.f. Bushberg, et al. The Essential Physics
29 30
of Medical Imaging, 2nd ed., p. 434. UW and Brent K. Stewart PhD, DABMP of Medical Imaging, 2nd ed., p. 433. UW and Brent K. Stewart PhD, DABMP

Gradient Recalled Echo (GRE) Acquisition Echo Planar Image (EPI) Acquisition

Similar to SE but with readout Extremely fast imaging


gradient reversal for 180
180 pulse Single (1 TR) and multi-
multi-shot
Repetition of acq. for each PE 90
90 flip, PEG/FEG, 180
180 flip
With small flip angles and gradient Oscillating PEG/FEG blips
blips
reversals large reduction in TR stimulate echo formation
and TE fast acq.
Rapid zig-
zig-zag
zag k-space filling
PEG rewinder pulse (opp. polarity)
Acq. occurs in a period < T2*:
to maintain relationship between
25-
25-50 msec
pulses (due to short TR)
High demands on sampling
Acq. time=TR
time=TR no. PE steps NEX
rate, gradient coils and RF
Example (256x192 matrix, deposition limitations (SAR)
TR=30): 15.5 sec
Poor SNR, low res. (642) and
SNR and artifacts; one slice many artifacts
GRASS, FISP, FLASH, etc. Real-
Real-time
time snapshot

c.f. Bushberg, et al. The Essential Physics c.f. Bushberg, et al. The Essential Physics
31 32
of Medical Imaging, 2nd ed., p. 434. UW and Brent K. Stewart PhD, DABMP of Medical Imaging, 2nd ed., p. 435. UW and Brent K. Stewart PhD, DABMP

UW and Brent K Stewart, PhD, DABMP 8


Magnetic Resonance Imaging (MRI) - Chapter 15 25 January, 8 and 15 February, and 1 March 2007

Spiral K-
K-space Acquisition Gradient Moment Nulling

Simultaneous oscillation of In SE and GRE SSG/FEG


PEG/FEG to sample data balanced so that the uniform
during echo formation in a dephasing caused by the initial
spiral starting at k-
k-space origin gradient application is
rephased by an opposite
Regridding to 2D k- k-space polarity gradient of equal area
array for 2D FT Moving spins phase
Efficient method placing dispersal not compensated
maximum samples in the low- low- Constant flow: spins can be
frequency area of k- k-space rephased with a gradient triplet
Like EPI sensitive to T2*: field Higher-
Higher-order corrections
inhomogeneities and Applied to both SSG/FEG to
susceptibility agents correct motion ghosting and
pulsatile flow
A = -1, B = 3 and C = -3

c.f. Bushberg, et al. The Essential Physics c.f. Bushberg, et al. The Essential Physics
33 34
of Medical Imaging, 2nd ed., p. 436. UW and Brent K. Stewart PhD, DABMP of Medical Imaging, 2nd ed., p. 437. UW and Brent K. Stewart PhD, DABMP

Raphex 2001 Diagnostic Questions 3D Fourier Transform Image Acquisition

D50. Which of the following does NOT generally affect


D50. Uses a broadband, non-
non-
the total exam time of an MRI study? selective RF pulse to excite a
large spin volume
Acq. time = TR no. PE steps
A. # of acquisitions (z) no. PE steps (y) NEX
B. # of frequency encoding steps SE: TR=600, 1283 164 min.
GRE: TR=50, 1283 14 min.
C. # of phase encoding steps
Isotropic or anisotropic (<time)
D. # of pulse sequences in the study High SNR thin slice recon.
E. TR prob. for motion artifacts
Volume for 3D slice/dice

c.f. Bushberg, et al. The Essential Physics


35 36
UW and Brent K. Stewart PhD, DABMP of Medical Imaging, 2nd ed., p. 438. UW and Brent K. Stewart PhD, DABMP

UW and Brent K Stewart, PhD, DABMP 9


Magnetic Resonance Imaging (MRI) - Chapter 15 25 January, 8 and 15 February, and 1 March 2007

Image Characteristics and Quality Spatial Resolution

Spatial Resolution and Contrast Sensitivity Dependent on


Signal-
Signal-to-
to-Noise Ratio (SNR) FOV: pixel size
Gradient strength: FOV
Basis for evaluating MR image characteristics
Receiver coil characteristics
Voxel Volume
Sampling bandwidth
Signal Averages (NEX) Image matrix: 1282 through 1024 x 512
RF Bandwidth In plane: 0.5-
0.5-1.0 mm (0.1-
(0.1-0.2 mm surface coil)
RF Coil Quality Factor Slice thickness: 5-
5-10 mm
Magnetic Field Strength Higher B0 (e.g., 3.0 T) larger SNR thinner slices
Cross Excitation However, RF heating, T1, T1 contrast and artifact
Image Acquisition and Reconstruction Algorithms

37 38
UW and Brent K. Stewart PhD, DABMP UW and Brent K. Stewart PhD, DABMP

Contrast Sensitivity Signal-


Signal-to-
to-Noise Ratio (SNR)

NEX
(, T1, T2, flow, pulse param.)
Major attribute of MR = f ( SNR = I Volvoxel f (QF ) f 2 ( B) f 3 (slice gap) f 4 (recon.)
BW 1
MR contrast agents, usually susceptibility agents (e.g.,
Gadolinium) disrupt local B field to enhance T2 decay or I = intrinsic signal intensity based on pulse sequence
provide additional relaxation mechanisms for T1 decay Volvoxel = voxel volume = f (FOV, matrix, slice thickness)
important enhancement agents for differentiation of
normal and diseased tissues NEX = number of excitations
Absolute contrast sensitivity of an MR image is ultimately BW = freq. BW of RF receiver
limited by the SNR and presence of image artifacts f1 (QF) = func. of coil quality factor param. (tuning coil)
f2 (B) = function of magnetic field strength
f3 (slice gap) = function of interslice gap effects
f4 (recon.) = function of reconstruction algorithm

39 40
UW and Brent K. Stewart PhD, DABMP UW and Brent K. Stewart PhD, DABMP

UW and Brent K Stewart, PhD, DABMP 10


Magnetic Resonance Imaging (MRI) - Chapter 15 25 January, 8 and 15 February, and 1 March 2007

Voxel Volume Signal Averages (NEX)

FOVx FOVy
Volvoxel = Slice thickness ( z ) SNR NEX
No. pixels, x No. pixels, y

SNR voxel volume Doubling SNR requires NEX = 4


matrix size or slice thickness SNR NEX < 1: or NEX
NEX: half Fourier imaging PE-
PE-matrix dimension + 1
NEX: PE-
PE-matrix dimension
Missing data synthesized from the k-
k-space matrix
SNR

41 42
UW and Brent K. Stewart PhD, DABMP UW and Brent K. Stewart PhD, DABMP

RF Bandwidth RF Coil Quality Factor

Range of freq. to which the RF Indication of RF coil sensitivity to induced currents in


detector is tuned response to signal emanating from the patient
Narrow BW SNR
1/
1/(BW)
Patient loading: electrical impedance characteristics of
1/T (dwell time time
BW = 1/ the body variation of B field, different for each patient
between FID sampling) Tuning the receiver coil to 0 mandatory
Narrow BW T noise Also dependent on volume of subject : coil volume
(SNR SQRT[
SQRT[T])
Body coil located in magnet bore: moderate QF
BW gradient strength
Surface coil: high QF
chem. shift artifacts)
Also requires longer sampling Trade-
Trade-off with FOV uniformity
time and affects TEmin which in Body coil: relatively uniform over FOV
(/2
BW = ( /2)GxFOVx
turn may affect num. slices/TR Surface coil: signal falls off abruptly (1/r3-5)
remember: /2
remember: =
/2
Hz-cm-1/mT-
426 Hz- /mT-m-1
c.f. Bushberg, et al. The Essential Physics
43 44
of Medical Imaging, 2nd ed., p. 441. UW and Brent K. Stewart PhD, DABMP UW and Brent K. Stewart PhD, DABMP

UW and Brent K Stewart, PhD, DABMP 11


Magnetic Resonance Imaging (MRI) - Chapter 15 25 January, 8 and 15 February, and 1 March 2007

Magnetic Field Strength Cross Excitation

Influences SNR: SNR B0 Due to non-


non-rectangular RF slice selection profiles
SNR about twice that at 3.0T than at 1.5T Overlap of adjacent slices in multislice sequence
due to T1 lengthening as B0 depending on TR
Saturates spins contrast
Other considerations mitigate SNR improvement Use interslice gaps or multislice interleaving
Longer T1
Greater RF absorption (and heating)

45 46
UW and Brent K. Stewart PhD, DABMP UW and Brent K. Stewart PhD, DABMP

Image Acquisition and Reconstruction Algorithms Raphex 2003 Diagnostic Questions

Profound effect on SNR D54. In MRI the signal-


D54. signal-to-
to-noise ratio can be increased by
Acquisition methods in order of increasing SNR: all of the following except:
Point
Line A. Decreasing the slice thickness
2DFT
B. Increasing the number of acquisitions
3DFT
C. Increasing the static magnetic field strength
Volume of tissue the major contributing factor in SNR
D. Increasing TR
High-
High-pass filtration methods SNR
E. Switching from a volume to a surface coil
Low-
Low-pass filtration methods SNR, but spat. resol.

47 48
UW and Brent K. Stewart PhD, DABMP UW and Brent K. Stewart PhD, DABMP

UW and Brent K Stewart, PhD, DABMP 12


Magnetic Resonance Imaging (MRI) - Chapter 15 25 January, 8 and 15 February, and 1 March 2007

Instrumentation Magnets

Magnets Magnet performance criteria:


Resistive Field strength
Temporal stability
Superconductive
Field homogeneity
Permanent
Air core magnets
Ancillary Equipment Wire wrapped cylinders
Magnet Siting and Shielding B0 produced through wire
current flow:
Quality Control
B0 parallel to core (usu. horiz.)
Solid core magnets
Permanent magnets
Wire wrapped iron core
B0 between poles (usu. vert.)

c.f. Bushberg, et al. The Essential Physics c.f. Bushberg, et al. The Essential Physics
49 50
UW and Brent K. Stewart PhD, DABMP of Medical Imaging, 2nd ed., p. 374. UW and Brent K. Stewart PhD, DABMP of Medical Imaging, 2nd ed., p. 458.

Permanent Magnets Superconductive Magnets

Ferromagnetic properties of Air core: 1m diam., 2-


2-3m depth
Fe, Ni, Co and alloys Wrapped with supercon. wire
Bulky and heavy, though new Liquid helium cooling
lighter alloys B0: 0.3-
0.3-3.0 T clinical (4-
(4-7 T
Finding a niche in clinical MRI research)
B0: 0.1-
0.1-0.35 T High field uniformity: <1 ppm
Lowest operating costs over 40 cm DSV
Field uniformity typically less Most widely used
than superconductive with Disadvantages: high initial
similar FOV capital and siting costs,
Inability to turn off field in an cryogen costs, difficulties
emergency! turning B off in emergency and
extensive fringe fields

c.f. Bushberg, et al. The Essential Physics


51 52
UW and Brent K. Stewart PhD, DABMP of Medical Imaging, 2nd ed., p. 459. UW and Brent K. Stewart PhD, DABMP

UW and Brent K Stewart, PhD, DABMP 13


Magnetic Resonance Imaging (MRI) - Chapter 15 25 January, 8 and 15 February, and 1 March 2007

Resistive Magnets Ancillary Equipment (1)

Either air core or solid core Shim coils active or passive,


Continuous electrical power ($) adjust B0 to homogeneity
Produces a significant amount Gradient coils noise caused
of heat (cooling system usu. by torque on coil and flexing
city or chilled water) RF coils transmitter and body
B0: 0.1-
0.1-0.7 T receiver within bore covers
Able to turn off magnet in an RF coils need to be tuned
tuned
emergency prior to each acquisition
Open design Kinds: bird-
bird-cage, single-
single-turn
Fringe field well contained solenoid, saddle, surface and
phased-
phased-array
Relatively poor
uniformity/homogeneity Quadrature detection SNR
by 2

c.f. http://homepage2.nifty.com/kirislab/chap5_mri/ c.f. Bushberg, et al. The Essential Physics


53 54
UW and Brent K. Stewart PhD, DABMP mri_images/imagingSystem/gradientCoilSet.gif UW and Brent K. Stewart PhD, DABMP of Medical Imaging, 2nd ed., p. 460.

Ancillary Equipment (2) Magnet Siting and Shielding

Pulse programmer Superconductive magnets:


Control interfaces extensive fringe fields
Patients w/ pacemakers or
RF transmitter
ferromagnetic aneurysm clips:
RF detector (coils) avoid fringe fields >0.5 mT (5g)
RF amplifiers Magnetically sensitive
Gradient power supplies equipment: video monitors,
cameras and fluoroscopic II
ADC electronics
Areas above 1.0 mT (10 g)
Computer system require controlled and
Image display restricted access w/ signs
Stray RF signal protection:
Faraday cage (copper
sheeting/mesh)

c.f. Bushberg, et al. The Essential Physics c.f. Bushberg, et al. The Essential Physics
55 56
of Medical Imaging, 2nd ed., p. 461. UW and Brent K. Stewart PhD, DABMP of Medical Imaging, 2nd ed., p. 463. UW and Brent K. Stewart PhD, DABMP

UW and Brent K Stewart, PhD, DABMP 14


Magnetic Resonance Imaging (MRI) - Chapter 15 25 January, 8 and 15 February, and 1 March 2007

Quality Control Raphex 2001 Diagnostic Questions

Periodical checking of: D45. Superconducting magnets, compared to resistive


Magnetic field strength magnets:
Magnetic field homogeneity
System field shimming
Gradient linearity A. Are less expensive
System RF tuning B. Are more easily turned off
Receiver coil optimization
Display monitors
C. Do not require liquid helium
ACR MRI accreditation prog. D. Have higher field strength
Uses phantoms composed of
materials that simulate patient
relaxation times

c.f. Bushberg, et al. The Essential Physics


57 58
of Medical Imaging, 2nd ed., p. 464. UW and Brent K. Stewart PhD, DABMP UW and Brent K. Stewart PhD, DABMP

Raphex 2003 Diagnostic Questions Safety and Bioeffects (1)

D57-D59. Match the following MRI terms. (Answers may be used


D57- Important safety considerations
more than once.) Strong magnetic fields
RF energy
A. Gradient fields Time-
Time-varying magnetic gradient fields
B. RF Confined imaging space (claustrophobia)
C. Shim coils Noisy operation
D. T1 Implants ferromagnetic (torque) and non-
non-ferromagnetic (heat)
E. T2 Ferromagnetic implements (IV pole, gas cylinders, etc.)
Long-
Long-term biological effects of high magnetic fields not
D57. Used to adjust magnetic field uniformity
D57.
well known
D58. Used to localize MR signal
D58.
D59. Used to tip the net magnetization of spins
D59. Most common bioeffect: tissue heating (RF/gradients)

59 60
UW and Brent K. Stewart PhD, DABMP UW and Brent K. Stewart PhD, DABMP

UW and Brent K Stewart, PhD, DABMP 15


Magnetic Resonance Imaging (MRI) - Chapter 15 25 January, 8 and 15 February, and 1 March 2007

Safety and Bioeffects (2) Safety and Bioeffects (3)

Static Magnetic Fields


> 20 T: membrane permeability, enzyme kinetic changes and altered
biopotentials
< 10 T: these effects have not been demonstrated
Varying Magnetic Field Effects
Gradient switching can cause current flow
At very high levels: visual phosphenes
Magnetic Field, RF Exposure and Noise Limits
RF exposure causes tissue heating
Power deposition limits: (< 1
1 C head, < 2
2 C trunk, < 3
3 C extremities)
4 W/kg averaged over the whole body for any 15- 15-minute period
3 W/kg averaged over the head for any 10-10-minute period; or
8 W/kg in any gram of tissue in the extremities for any period of
of 5 min

c.f. Bushberg, et al. The Essential Physics


61 62
of Medical Imaging, 2nd ed., p. 466. UW and Brent K. Stewart PhD, DABMP UW and Brent K. Stewart PhD, DABMP

Raphex 2000 Diagnostic Questions Artifacts

D49. Patients who have MRI scans should be screened


D49. Machine Dependent Artifacts
to eliminate those who have: Susceptibility Artifacts
Gradient Field Artifacts
A. Internal steel fragments Radiofrequency Coil Artifacts
B. Metallic prostheses Radiofrequency Artifacts
C. Pacemakers K-space Errors
D. Surgical clips Chemical Shift Artifacts
E. All of the above Ringing Artifacts
Wraparound Artifacts
Partial Volume Artifacts

63 64
UW and Brent K. Stewart PhD, DABMP UW and Brent K. Stewart PhD, DABMP

UW and Brent K Stewart, PhD, DABMP 16


Magnetic Resonance Imaging (MRI) - Chapter 15 25 January, 8 and 15 February, and 1 March 2007

Machine Dependent Artifacts Susceptibility Artifacts

Magnetic susceptibility: ratio of


Magnetic field inhomogeneities distortion or
induced internal magnetization in
misplacement of anatomy a tissue to external magnetic field
(B0)
Proper site planning, self-
self-shielded magnets, automatic Drastic changes in mag. suscept.
distort B0
shimming and preventative maintenance procedures Tissue-
Tissue-air interfaces: lungs and
> homogeneity sinuses rapid T2*
Metal: ferrous or not
Focal field inhomogeneities ferromagnetic objects: Paramagnetic agents (Gd)
field distortions, signal void Paramagnetic effects shorten T2
Hydration layer interactions
shorten T1
Mag. suscept. of blood
degradation products
Diagnose the age of a
hemorrhage
EPI diffusion study suffers from severe susceptibility
artifact due to retained metal after surgery. Courtesy,
GE Medical Systems.

65 66
UW and Brent K. Stewart PhD, DABMP UW and Brent K. Stewart PhD, DABMP

Gradient Field Artifacts Radiofrequency Artifacts

Reconstruction algorithm Stray RF signals


assume linear gradients TV, radio, electric motors,
Tendency for gradient field fluorescent lights & computers
strength at periphery of FOV to Narrowband: zipper artifact
deviate from linear assumption perpendicular to FEG direction
Reduce FOV or lower gradient Broadband: herringbone
artifact across larger area
strength
RF shielding : Faraday cage
Need balanced gradient
strength for PEG and FEG RF quadrature coils:
Otherwise non-
non-square pixels imbalanced amplifiers DC
offset
Causes ghosting of objects
diagonally in image The scanner room door was left open during the
acquisition causing the zipper artifacts shown. c.f.,
www.spectroscopynow.com/Spy/pdfs/mritutor.pdf

c.f. Bushberg, et al. The Essential Physics 67 68


of Medical Imaging, 2nd ed., p. 449. UW and Brent K. Stewart PhD, DABMP UW and Brent K. Stewart PhD, DABMP

UW and Brent K Stewart, PhD, DABMP 17


Magnetic Resonance Imaging (MRI) - Chapter 15 25 January, 8 and 15 February, and 1 March 2007

Radiofrequency Coil Artifacts K-space Error Artifacts

Surface coils variations in Artifactual superimposition of wave patterns across the FOV
uniformity across the image Even one bad pixel can produce a significant artifact, especially
especially
caused by RF attenuation, RF when at or near k-
k-space DC data point (center)
mismatching and sensitivity
falloff with distance
Non-
Non-rectangular RF pulses:
slice-
slice-to-
to-slice interference
T2-
T2-weighted SNR
T1-
T1-weighted image
contrast
Interslice gaps and pseudo-
pseudo-
rectangular RF pulses
Slice interleaving

c.f. Bushberg, et al. The Essential Physics c.f. Bushberg, et al. The Essential Physics
69 70
of Medical Imaging, 2nd ed., p. 450. UW and Brent K. Stewart PhD, DABMP of Medical Imaging, 2nd ed., p. 451. UW and Brent K. Stewart PhD, DABMP

Motion Artifacts Chemical Shift Artifacts


f0 variations resulting from
Mostly occur along the PEG intrinsic magnetic shielding
direction ghost images f = (1 - ) (/2
/2) B0
Compensation methods: Distinct peaks in MR spectrum
Cardiac/respiratory gating Fat: =3.5 ppm lower than H20
Respiratory ordering
B chemical shift
Signal averaging
G chemical shift
Short TE T1-
T1-w SE sequences
Cannot distinguish freq. shift
Gradient moment nulling
by FEG or chemical shift
Presaturation pulses applied
outside the imaging region Misregistration of H20 and fat
moieties anatomical shift
Cure: G, but SNR
Cure: off-
off-reson. presat. pulse
Cure: STIR bounce point
c.f. Bushberg, et al. The Essential Physics c.f. Bushberg, et al. The Essential Physics
71 72
of Medical Imaging, 2nd ed., p. 451-
451-2. UW and Brent K. Stewart PhD, DABMP of Medical Imaging, 2nd ed., p. 454. UW and Brent K. Stewart PhD, DABMP

UW and Brent K Stewart, PhD, DABMP 18


Magnetic Resonance Imaging (MRI) - Chapter 15 25 January, 8 and 15 February, and 1 March 2007

Ringing Artifacts Wraparound (


(Aliasing
Aliasing) Artifacts

AKA Gibbs phenomenon


phenomenon Result of mismapping anatomy
Occurs near sharp boundaries that lies outside the FOV, but
and high-
high-contrast transitions within the slice volume
Multiple, regularly spaced Caused by:
parallel bands of alternating Non-
Non-linear gradients
bright/dark signal fading with Undersampling of frequencies
distance within the signal envelope
(Nyquist sampling limit)
Lack of high-
high-frequency signals
causes ringing
ringing at sharp FT cannot distinguish freq. >
transitions Nyquist limit lower freq.
Most likely for small matrix Cure: low-
low-pass filter (<Nyquist)
dimensions Cure: FOV G or BW
Skull/brain interface Cure: number of PE steps
(/2
BW = ( /2)GFOVx = 1/
1/T
c.f. Bushberg, et al. The Essential Physics c.f.,
c.f. www.spectroscopynow.com/ c.f. Bushberg, et al. The Essential Physics
73 74
of Medical Imaging, 2nd ed., p. 456. UW and Brent K. Stewart PhD, DABMP Spy/pdfs/mritutor.pdf of Medical Imaging, 2nd ed., p. 441 & 457. UW and Brent K. Stewart PhD, DABMP

Partial Volume Artifacts Raphex 2002 Diagnostic Questions

Due to finite voxel dimensions D57. All of the following are MRI artifacts except:
D57.
Cure: pixel size/slice thickness
Problem: SNR for similar imaging time A. Chemical shift
B. Ring
C. Susceptibility
D. Wrap-
Wrap-around
E. Zipper

75 76
UW and Brent K. Stewart PhD, DABMP UW and Brent K. Stewart PhD, DABMP

UW and Brent K Stewart, PhD, DABMP 19


Magnetic Resonance Imaging (MRI) - Chapter 15 25 January, 8 and 15 February, and 1 March 2007

Some Advanced Topics Signal from Flow (1)

Time-
Time-of-
of-Flight (TOF) MR Angiography (MRA) The MR signal from moving fluids (vascular and CSF) is
Phase Contrast MRA complicated by many factors:
Flow velocity
Magnetization Transfer Contrast (MTC) Vessel orientation
Perfusion and Diffusion Contrast Laminar vs. turbulent flow patterns
fMRI and BOLD Imaging Pulse sequences
Image acquisition modes
Flow related mechanisms combine with image
acquisition parameters to alter contrast
Bright-
Bright-blood
blood to black-
black-blood
blood
Can be a source of artifacts
Exploited to produce MR angiography images

77 78
UW and Brent K. Stewart PhD, DABMP UW and Brent K. Stewart PhD, DABMP

Signal from Flow (2) Signal from Flow (3)

Flow-
Flow-related enhancement Low signal intensities: high-
high-
Even-
Even-echo rephasing velocity signal loss
(prominent in slow laminar Nuclei move out of slice during
flow veins) echo reformation (nothing
Gradient echo images focused in Mxy plane no or
blood):
(unsaturated blood): little FID signal)
signal)
velocity, slice thinness
thinness and
TR

c.f. Hashemi, et al. MRI - c.f. Hashemi, et al. MRI -


79 80
The Basics, 2nd ed., p. 300. UW and Brent K. Stewart PhD, DABMP The Basics, 2nd ed., p. 296. UW and Brent K. Stewart PhD, DABMP

UW and Brent K Stewart, PhD, DABMP 20


Magnetic Resonance Imaging (MRI) - Chapter 15 25 January, 8 and 15 February, and 1 March 2007

Signal from Flow (4) Time-


Time-of-
of-Flight - TOF - MR Angiography MRA (1)

Signal from blood dependent Single- (=45-


Single-slice GRE ( =45-60
60, TR=50 msec, TE=few
on relative saturation of tissues msec)
and the incoming blood flow
Unsaturated spins entering the Differentiates moving blood (unsaturated) from stationary
imaged slice(s) large FIDs tissues (saturated)
In some cases blood signal Penetration of unsaturated blood depends on: velocity
eliminated through pre-
pre- (magnitude and direction)
saturation pulses outside of
imaged slice(s) 2D stack of slices usually acquired
Black blood
blood (flow void) also Blood moving in unwanted direction (e.g., arterial and
caused by rapidly flowing and venous) is eliminated with a presaturation pulse in an
turbulent blood (no full 180
180 adjacent slice
pulse)

c.f. Bushberg, et al. The Essential Physics


81 82
of Medical Imaging, 2nd ed., p. 443. UW and Brent K. Stewart PhD, DABMP UW and Brent K. Stewart PhD, DABMP

Time-
Time-of-
of-Flight - TOF - MR Angiography MRA (1) Phase Contrast MR Angiography MRA (1)

GRE technique provides Relies on phase change ( ())


poor anatomic contrast, for moving protons (blood);
but a high-
high-contrast = /2
/2 Gx vx t2
bright blood
blood signal
Application of + and then
Maximum intensity
polarity gradients in rapid
projection (MIP) along
specific viewing angles succession ( (T)
used to generate a Second acquisition during
series of images for same phase encode cancels
display ()
) for stationary spins
TOF MRA often Moving spins accumulate ( ())
produces variation in
vessel intensity Amount of (()) (T) and v
dependent on orientation
wrt. viewing plane

c.f. Bushberg, et al. The Essential Physics c.f. Bushberg, et al. The Essential Physics
83 84
of Medical Imaging, 2nd ed., p. 444. UW and Brent K. Stewart PhD, DABMP of Medical Imaging, 2nd ed., p. 446. UW and Brent K. Stewart PhD, DABMP

UW and Brent K Stewart, PhD, DABMP 21


Magnetic Resonance Imaging (MRI) - Chapter 15 25 January, 8 and 15 February, and 1 March 2007

Phase Contrast MR Angiography MRA (2) fMRI and BOLD Imaging (1)

Intensity variations depend on BOLD (B(Blood Oxygen Level-


evel-Dependent)
())
the amount of (
Differential contrast generated by blood metabolism in
Brightest pixels highest +v,
mid-
mid-gray 0v, lowest v
brain
Unlike TOF MRA, the phase Oxyhemoglobin deoxyhemoglobin (paramagnetic)
contrast image is inherently increases magnetic susceptibility and induced signal loss
quantitative (increased T2*)
When calibrated provides an
estimate of the mean blood
flow v (magnitude and
direction)
2D and 3D possible

c.f. Bushberg, et al. The Essential Physics


85 86
of Medical Imaging, 2nd ed., p. 446. UW and Brent K. Stewart PhD, DABMP UW and Brent K. Stewart PhD, DABMP

fMRI and BOLD Imaging (2) Perfusion and Diffusion Contrast (1)

Areas of metabolic activity Perfusion of cells via capillary bed


correlated signal (functional
MR) Exogenous tracer methods
2H, 3He, 17O and 19F experimental procedures
Subtract post-
post-stimulus image
from pre-
pre-stimulus image Intravascular blood-
blood-pool agents: Gd-
Gd-DTPA
Color-
Color-coded overlay to a Endogenous tracer methods
grayscale anatomic image Labeling of inflowing spins (
(black blood
blood): tagging
demonstrate activity(t)
Tagged spins perfuse into tissues MR signal intensity
correlating with stimulus(t)

87 88
UW and Brent K. Stewart PhD, DABMP Images courtesy of Stanford University UW and Brent K. Stewart PhD, DABMP

UW and Brent K Stewart, PhD, DABMP 22


Magnetic Resonance Imaging (MRI) - Chapter 15 25 January, 8 and 15 February, and 1 March 2007

Perfusion and Diffusion Contrast (2) Perfusion and Diffusion Contrast (3)

Diffusion depends on the random motion of H2O In vivo structural integrity of


tissues measured apparent
molecules in tissues diffusion coefficient maps
Interactions of the local cellular structure with the Sensitive indicator for early
diffusing H2O molecules produces anisotropic, detection of
Spine and spinal cord
directionally dependent diffusion

pathophysiology
Diffusion-
Diffusion-weighted sequences use a strong gradient Ischemic injury

signal differences based on mobility/directionality Spin-


Spin-echo and echoplanar pulse
sequences with diffusion gradients
Tissues with H2O mobility have greater signal loss Obstacles
Diffusion-
Diffusion-weighted image (DWI) with gray
scale-
scale-encoded diffusion coefficients.
Sensitivity to head/brain motion
Eddy currents

c.f. Bushberg, et al. The Essential Physics


89 90
UW and Brent K. Stewart PhD, DABMP of Medical Imaging, 2nd ed., p. 410. UW and Brent K. Stewart PhD, DABMP

Magnetization Transfer Contrast (1) Magnetization Transfer Contrast (2)

Result of selective observation This process affects only those


of the interaction between the p+ having chemical exchange
p+ in free H2O molecules and with the macromolecules and
p+ in macromolecular proteins improves image contrast
due to coupling or chemical Anatomic imaging of heart,
exchange eye, MS, knee cartilage and
general MR angiography
Can be excited separately
Tissue characterization
using narrow-
narrow-band RF possible as the magnetization
Magnetization transferred from transfer ratio (MTCon/MTCoff) is
macromolecular p+ to free H2O caused in part by tissue-
tissue-
p+ specific surface chemistry
Reduced signal from adjacent MR arthrograms of shoulder in 32-year-old man with suspected gleno-
humeral instability. Axial 3D gradient-echo MR image obtained using
free H2O p+ parametric magnetization transfer pulses no discernible magnetization
transfer contrast in injected fluid or in fatty marrow spaces, whereas
degree of magnetization transfer contrast varies in skeletal muscle,
cartilage, and capsular supporting structures (color scale = 0-100%).
c.f. Yao L, Thomasson D. Magnetization transfer contrast
c.f. Bushberg, et al. The Essential Physics in rapid three-
three-dimensional MR imaging using segmented
91 92
of Medical Imaging, 2nd ed., p. 412. UW and Brent K. Stewart PhD, DABMP radiofrequency prepulses. AJR 2002; 179: 863-
863-5 . UW and Brent K. Stewart PhD, DABMP

UW and Brent K Stewart, PhD, DABMP 23