Basic M R I Imaging Methods

1
Basic M R I Imaging Methods

• Jerry Allison, Ph.D.,
• and Nathan Yanasak, Ph.D.

2
Outline I. Review of Spin Echo Imaging II.
Inversion Recovery (IR) III. Gradient Recalled
Echo IV. Three Dimensional (Volume)
Techniques V. Fast Imaging Techniques VI.
Echoplanar Imaging Other techniques not
necessarily covered in this lecture (but in
others) Parallel imaging, flow-related imaging
techniques
3
Spin Echo Imaging

• The Spin Echo imaging technique has the advantage
  that it is not as sensitive to static
  inhomogeneity of the magnet and inhomogeneity
  caused by magnetic susceptibility of patient
  tissue.
• Relaxation caused by flow, diffusion or proton
  exchange are not compensated by the spin echo
  technique.

3
4
Hahn Spin Echo

• The concept of Spin Echo production was
  developed by Hahn. Spin echoes are sometimes
  referred to as Hahn spin echoes.
• The Hahn Spin Echo technique consists of a 90o
  flip, followed by an interpulse delay time, t, a
  180o flip, followed by a second interpulse delay
  (t). A spin echo occurs at echo time TE (2t)
  following the initial 90o flip.

4
5
Hahn Spin Echo (continued)

• The Hahn Spin Echo was developed in 1950 for use
  in nMR long before the advent of MRI. The Hahn
  Spin Echo was used for the measurement of T2
  values.

5
6
Spin Echo Pulse Sequence

• By applying repeated 180o RF pulses, multiple
  echoes can be generated for imaging at different
  echo times for different image contrast, in
  addition to measurement of T2 values.

6
7
Spin Echo Imaging (continued)

• Remember, the value of the repetition time (TR)
  and the echo time (TE) can be varied to control
  contrast in spin echo imaging.
• For general spin echo, TR is used to mediate T1
  contrast, and TE is used for T2 contrast.
• Long TR, short TE ? Proton Density Weighting
• (want to minimize both T2, T1 contrast)
• Long TR, medium TE ? T2 Weighting
• Medium TR, short TE ? T1 Weighting

7
8
Spin Echo Images
T2 weighted
PD weighted
T1 weighted
TR 510 TE 14 2min 7sec for 17 slices
TR 4500 TE 15eff (ETL7) 2min 39sec for 24 slices
TR 4500 TE 105eff (ETL7) 2min 39sec for 24 slices
8
9
Inversion Recovery (IR)

• Inversion recovery pulse sequences are useful
  for
• Creation of heavily T1-weighted images without a
  dominant contribution from fat (e.g. brain, liver
  and musculoskeletal imaging).
• Suppression of selected tissues (e.g. orbital
  fat, liver screening, fatty tumors, CSF)

9
10
Inversion Recovery (IR) (continued)

• The basic IR pulse sequence consists of a 180o
  inversion pulse inserted before whatever
  sequence you usually choose for a particular
  contrast. The standard 90o RF excitation pulse
  of your sequence follows the inversion pulse
  after an inversion time TI.

The remainder of whatever sequence that you
choose goes here.
10
11
Inversion Recovery (IR) (continued)

• First, application of the 180o RF pulse inverts
  the macroscopic magnetization. During the
  inversion time, the macroscopic magnetization
  shrinks along the negative Z axis, eventually
  passes through Z 0 and regrows along the
  positive Z axis toward thermal equilibrium.
  Before the macroscopic magnetization is fully
  relaxed, the 90o RF pulse flips the partially
  relaxed longitudinal magnetization into the
  transverse plane in order to measure the signal
  induced in an RF coil.

11
12
Consider two voxels, one of fat and one of H2O
This method of fat suppression is sometimes
called short TI inversion recovery or STIR
imaging.
12
13
Animation of Inversion Process for STIR
Transverse magnetization for shorter T1 tissue is
suppressed.
Transverse magnetization for longer T1 tissue is
retained.
13
14
Inversion Recovery (IR) (continued)

• In many inversion recovery imaging sequences, the
  90o pulse is followed by a 180o pulse in order
  to produce a spin echo at time TE following the
  90o pulse.

14
15
15
16
IR Image Comparison
FLAIR fluid attenuated IR (T2-weighted spin echo)
STD T2 weighting
vs
FLAIR Inversion time 2.5sec (CSF is
suppressed) TR 10sec TE 119msec (ETL7) 3min
49sec for 19 slices
16
17
IR Image Comparison
vs
STIR short tau inversion recovery
STD T1 weighting
STIR image demonstrating high signal surrounding
the distal radius fracture indicating marrow
edema and hemorrhage (McAlinden, P.S., et al.,
Imaging 2003 15 180-192). STIR is an inversion
recovery sequence used to null fat.
17
18
Gradient Recalled Echo

• Gradient recalled echo techniques have great
  versatility. Gradient recalled echo techniques
  (GRE), or field echo techniques, are similar to
  spin echo techniques. GRE techniques include
  GRASS, SPGR, FLASH, FISP, PSIF and many, many
  others.
• Advantage A variety of contrasts can be produced
  while imaging rapidly.

18
19
Gradient Recalled Echo (continued)

• GRE differences from spin echo
• 1. The creation of the echo is accomplished
  solely by gradient magnetic fields.
• 2. Since 180o RF pulses are not used to create
  the echo, the deposition of RF energy in the
  patient is lower (less heating of patient
  tissues).
• 3. Static inhomogeneity of the magnet and
  inhomogeneity caused by magnetic susceptibility
  of patient tissue are NOT corrected by gradient
  recalled echo techniques.
• T2 contrast strictly becomes T2 contrast with
  GRE.

19
20
Gradient Recalled Echo (continued)

• 4. The initial flip angle is frequently chosen
  to be less than 90o . The flip angle in gradient
  recalled echo techniques is called a . Selection
  of a as less than 90o prevents severe saturation
  of the spin population when TR is short for rapid
  imaging. The optimum value of a for a particular
  TR and tissue having spin lattice relaxation T1
  is called the Ernst angle.

e
-TR
cos(ae)
20
21
Gradient Recalled Echo (continued)
e
-TR
cos(ae)

• a is 68o for TR T1
• a is 38o for TR 0.25 T1
• a is 89o for TR 4 T1
• if TR 200 ms tissues having T1 800ms
• T1 800 ms contribute more signal than
• a 38o those having 800 gt T1 lt 800

21
22
Gradient Recalled Echo (continued)

• 5. A small flip angle means that only a small
  portion of the longitudinal magnetization is
  converted to transverse magnetization by each RF
  pulse. TR can be reduced substantially without
  resulting in severe saturation, reducing scan
  times. In comparison to RF spin echo (SE)
  techniques, the transverse magnetization measured
  is much smaller for GRE, but it is measured more
  frequently. Individual GRE images appear noisier
  than SE images, but the SNR per unit time is
  actually higher.

22
23
Gradient Recalled Echo (continued)

• 6. 3D or volume imaging can be accomplished
  (resulting in thinner slices).

23
24
Gradient Recalled Echo (continued)
and echo is formed by switching the polarity
during the subsequent application of the
frequency- encode gradient.
Unlike the spin-echo characteristics, FID decay
is induced by the frequency-encode prephasing
24
25
Movie showing the generation of echoes using
gradients
25
26
Gradient Recalled Echo Images
2D-FLASH TR 25msec TE 9msec a 35o 5.7sec per
slice
MIP (Maximum Intensity
Projection)
26
27
Gradient Recalled Echo Image
Multi Planar GRASS mixed T1/T2 weighting TR
500msec TE 13msec 2NEX a60o 3min 14 sec
for 15 slices
27
28
Three Dimensional Volume Techniques

• The short TRs of GRE pulse sequences enable the
  acquisition of volume or 3D data sets and utilize
  3DFFT reconstruction techniques.

28
29
Three Dimensional Volume Techniques

• Differences between 2DFFT and 3DFFT are as
  follows
• 1. The slice select gradient is used twice in
  3D techniques. Initially, the slice select
  gradient is turned on during the RF pulse to
  selectively excite the slab of tissue that
  comprises all of the slices.

29
30
Three Dimensional Volume Techniques (continued)

• 1. (continued)
• Subsequently, the slice select gradient plays
  out an ensemble of slice encoding gradients to
  encode each slice in the 3D data set with a
  unique set of phases and frequencies.

30
31
Comparison between 2D and 3D GRE Sequences
31
32
Three Dimensional Volume Techniques (continued)

• 2. Data is reconstructed along all 3 coordinate
  axes using a 3DFFT.
• 3. Image acquisition is longer than for
  acquisition of a single slice.
• Acquisition time is calculated as follows
• Ts phase encodes x TR x acquisitions
  x slices

32
33
Three Dimensional Volume Techniques (continued)

• For example A 3D MPRAGE technique to produce
  T1 weighting could be prescribed as
• TR 11.4msec
• TE 4.2msec
• a 12o
• slices 128 (1.4mm thick)
• NEX 1
• Matrix 128 x 192x 256
• Acquisition Time 6min 55sec

33
34
Three Dimensional Volume Techniques (continued)

• 4. The 3D techniques allow for thinner slices
  (less than 1mm ) than 2D techniques. Slices are
  contiguous without concern for crosstalk.
• 5. The Signal to Noise Ratio for a 3D slice is
  higher than for a 2D slice of comparable
  thickness.

34
35
Three Dimensional Volume Techniques (continued)

• 6. 3D voxels are isotropic (or nearly
  isotropic). The voxels are the same size in all
  3 dimensions. The dimensions of a typical 3D
  voxel are
• 1 mm x 1 mm x 1 mm. The acquisition of
  isotropic voxels enables the data set to be
  reformatted into any oblique plane without
  significant loss of resolution using Post
  Processing Techniques.

35
36
Three Dimensional Volume Image
MPRAGE Magnetization Prepared Rapid Gradient
Echo TR 11.4msec TE 4.2msec a12o 1.4mm 6min
55sec for 120 slices (168mm slab) Uses Inversion
Recovery
36
37
Fast Imaging Techniques

• There are several motivations for using fast
  imaging techniques
• 1. Motion artifact reduction
• 2. Dynamic imaging
• - dynamic contrast perfusion studies
• - kinematic studies
• 3. Reduced acquisition time without significant
  loss of image resolution or contrast.

37
38
Fast Imaging Techniques (continued)

• Fast techniques involving k-space
• Simple concept Each phase-encoding step in a
  sequence takes a certain amount of time to
  execute. If we reduce the number of
  phase-encoding steps, we reduce the scan time.
• Methods using reduced number of phase-encodings,
  and their tradeoffs
• Partial Fourier Phase Encoding lower SNR
• Reduced matrix lower spatial resolution
• Rectangular FOV smaller field-of-view

38
39
Fast Imaging Techniques (continued)

• Partial Fourier Phase Encoding (Fractional NEX
  Half NEX Half Fourier)
• Due to the symmetrical nature of k-space,
  slightly more than half of k-space can be
  acquired with the remaining data compensated by
  mathematical processes (symmetric in both
  directions). If half of k-space is filled (NEX
  0.5) the scan takes about half as long to acquire
  but has a lower signal-to-noise ratio.

39
40
Fast Imaging Techniques (continued)

• As a first approximation, lets replace about
  half of k-space with zeroes and reconstruct an
  image.

data
zeroes
40
41
Fast Imaging Techniques (continued)

• Compare ½ NEX with a full k-space scan (NEX1).

41
42
Fast Imaging Techniques (continued)

• Reduced matrix
• Eliminating the outer lines of k-space
  results in reduced scan time at the cost of
  coarser spatial resolution (larger voxels)

zeroes
42
43
Fast Imaging Techniques (continued)

• Rectangular FOV
• Covering k-space with fewer lines (larger
  phase encoding steps) reduces scan time, reduces
  field of view without loss of spatial resolution.

Smaller FOV creates aliasing in this case.
43
44
Fast Imaging Techniques (continued)

• Fast Spin Echo Techniques
• Fast spin echo techniques combine k-space
  reordering with production of multiple spin
  echoes within a single TR to reduce acquisition
  times without significant loss in image quality.
• Conventional spin echo techniques acquire one
  line of k-space during each repetition (TR).
  Fast spin echo techniques acquire 2 to 16 lines
  of k-space during each repetition.

44
45
Fast Spin Echo Techniques (continued)

• The initial 90o RF excitation pulse in a fast
  spin echo technique is followed by a series of 2
  to 16 180o RF pulses. This series of 180o RF
  pulses is referred to as the Echo Train and the
  number of 180o RF pulses in the series is
  referred to as the Echo Train Length. Each 180o
  RF pulse produces a spin echo which is acquired
  with a different phase encode gradient.

45
46
Fast Spin Echo Techniques (continued)

• The tradeoff is that, for each echo train,
  different lines in k-space will be observed at
  slightly different echo times. This leads to a
  slightly different T2 contrast for each echo in
  an echo train.
• Solution 1 scan middle lines of k-space with
  the first echo of the train, and outer lines with
  the latter echoes.
• Because the middle of k-space is responsible for
  image contrast and the outside of k-space
  produces the details, this approach preserves the
  overall desired contrast.
• Solution 2 scan some of the outer lines during
  the early echoes and others during the late
  echoes.

46
47
Fast Spin Echo Techniques (continued)
47
48
Fast Spin Echo Techniques (continued)

• Acquisition time for a fast spin echo sequence
  can be calculated as follows
• TR x phase encodes x
  acquisitions
• Echo Train Length
• Consider a conventional spin echo sequence
• TR 2000 msec
• phase encodes 256
• acquisitions 2
• ETL 1 ( one line of k-space per TR)

Ts
48
49
Fast Spin Echo Techniques (continued)

• Ts (2 sec) (256) (2) 17.07minutes
• (1)
• Now consider a fast spin echo sequence with
  identical parameters except an echo train length
  of 7
• Ts (2 sec) (256) (2) 2.44minutes
• (7)

49
50
Fast Spin Echo Image
TR 4500 TE 96eff Echo train length 7 2min 3sec
50
51
Echo Planar Techniques
EPI

• EPI techniques can acquire an image in 50-100
  msec. They are particularly useful for imaging of
  diffusion and perfusion and are also used in
  mapping brain function (fMRI).

51
52
Echo Planar Techniques

• Echo planar techniques (also known as single shot
  techniques) enable reconstruction of an image
  following a single RF excitation.
• Following an RF excitation, a long echo train of
  RF-generated spin echoes or gradient recalled
  echoes, each with different phase encoding, are
  acquired.

52
53
Echo Planar Techniques (continued)

• Most echo planar techniques acquire either
  64 or 128 echoes per RF excitation. Using Half
  Fourier (Half NEX) techniques, images having
  resolution of 128 x 256 or 256 x 256 can be
  reconstructed.

53
54
EPI Images
EPI images from a functional MRI dataset TR 5sec
TE 66msec 22 slices 90 acquisitions 1980 images
acquired in 7.5minutes
54
55
EPI Images
Diffusion gradients off
Diffusion gradients on
EPI images from a diffusion dataset TR 4sec TE
100msec 20 slices 140 images (7 diffusion
conditions) acquired in 80sec
55
56
EPI techniques require strong, fast gradients.
56
57
Echo Planar Techniques (continued)
Use of sinusoidal phase encode gradients produces
non-uniform (zig-zag) sampling of k-space. The
rectangular k-space matrix is interpolated from
non-uniform sampling, before applying a 2D FFT.
57
58
PARALLEL MRI

• Accelerates MRI!
• Multiple RF detector coils (Arrays)
• Multiple receivers (4, 8, 16)
• Supplements encoding provided by gradients
• Reduces constraints on gradient switching rates
• Reduces SAR issues
• Use of multiple detectors replaces need for some
  gradient pulses and RF pulses
• Multiplies imaging speed without increasing
  gradient switching rates or SAR

58
59
PARALLEL MRI

• SMASH (1997)
• SiMultaneous Acquisition of Spatial Harmonics
• Linear combinations of component coil signals to
  emulate directly the effects of encoding
  gradients
• SENSE (1999)
• SENSitivity Encoding
• Arbitrary k-space trajectories
• See http//www.mr.ethz.ch/sense/ for details

59
60
PARALLEL MRI

• Reduction of SAR is going to become more of a
  clinical issue as higher field MRI becomes more
  common. So, parallel MRI will become a fairly
  common technique (and probably a necessity for
  some protocols) in the near future.
• We will cover how this works in a separate
  lecture.

60