Magnetic field gradients
Spatial encoding relies on successively applying magnetic field gradients.First of all, a slice selection gradient (GSS) is used to select the anatomical volume of interest. Within this volume, the position of each point will be encoded vertically and horizontally by applying a phase encoding gradient (GPE), and a frequency-encoding gradient (GFE).
The different gradients used to perform spatial localization have identical properties but are applied at distinct moments and in different directions. Gradient equivalence in the three directions of space means that slices can be selected on any spatial plane. We’ll use the example of an axial plane to explain spatial encoding.
The gradient is symbolized by two triangles joined at the apex. This indicates that the gradient is bipolar with a positive component (added to B0) and a negative component (subtracted from B0).
The following diagram summarizes an MRI sequence of the Spin Echo type (figure 4.2), which will serve as a protoype to explain slice selection gradients, phase encoding and frequency encoding.
Selecting the slice plane: selective excitation
The first step of spatial encoding consists in
selecting the slice plane. To do this, a magnetic field gradient, the Slice
Selection Gradient (GSS), is applied perpendicular to the desired slice plane.
This is added to B0, and the protons present a resonance frequency variation
proportionate to GSS (Larmor equation). An RF wave is simultaneously applied,
with the same frequency as that of the protons in the desired slice plane. This
causes a shift in the magnetization of only the protons on this plane. As none
of the hydrogen nuclei located outside the slice plane are excited, they will
not emit a signal. The RF wave associated with the slice selection gradient and
the adapted resonance frequency, is called the selective pulse.
These protons located in the slice plane will
again be stimulated by the magnetic field gradients to encode their position in
horizontal and vertical directions.
Thickness, slice profile and selective RF pulse
An RF pulse does not have one frequency only (for this, it would need to be of infinite duration). It covers a certain bandwith, which depends on the shape of the pulse and its duration.The thickness of the slice can be varied by adjusting the bandwidth of the selective pulse and the amplitude of the slice selection gradient:
- For a fixed amplitude gradient, the wider the bandwidth, the greater the number of protons excited and the thicker the slice
- For a fixed bandwidth, the stronger the gradient, the greater the variation of precession frequency in space and the thinner the slice
Additional slice selection gradient lobe(s)
During selective pulse delivery, the magnetization shift gives rise to transversal magnetization, which will be subjected to the slice selection gradient.In the case of an excitation pulse at an angle below 180°, the slice selection gradient will have a spin dephasing effect due to the dispersion in the resonance frequency produced. To neutralize this effect, after applying the selective RF pulse (concomitant with the gradient) another gradient lobe is applied, along the same axis but in the opposite direction and with a surface (amplitude x time) equal to half the initial gradient lobe.
In the case of a 180° pulse, the dephasing effects neutralize symmetrically in relation to the centre of the RF pulse, so no rephasing lobe needs to be applied.
On the other hand, as the slice profile is imperfect, a 180° rephasing pulse, which will simply invert the magnetization of the excited spins, will also cause a shift in the undesired spins on the edge of the slice. To avoid this phenomenon, two identical gradient lobes can be added on each side of the 180° pulse:
- These lobes will have a symmetrical effect on the existing magnetization, before and after the 180° pulse, and will neutralize each other
- Magnetization on the edge of the slice created by the 180° pulse will be offset by the effect of the second added gradient lobe, which will cancel it out.
Phase encoding
The second step in spatial encoding consists in applying a phase encoding gradient, which we will choose to apply in the vertical direction. The phase encoding gradient (GPE) intervenes for a limited time period. While it is applied, it modifies the spin resonance frequencies, inducing dephasing, which persists after the gradient is interrupted. This results in all the protons precessing in the same frequency but in different phases. The protons in the same row, perpendicular to the gradient direction, will all have the same phase. This phase difference lasts until the signal is recorded.
On receiving the signal, each row of protons will be slightly out of phase. This translates as their signals being more or less out of phase.
To obtain an image, it is necessary to multiply the different dephased acquisitions, which are regularly incremented. For a spin echo sequence with « n » rows, we make « n » acquisitions each with a different phase encoding gradient.
Duration of a 2D imaging sequence
Duration = TR ∙ NPy ∙ NexWith
- TR = Repetition time
- NPy = Number of phase encoding steps
- Nex= Number of excitations
Frequency encoding
The final step in spatial encoding consists in applying a frequency encoding gradient, when the signal is received, in the last direction (horizontal in our example). This modifies the Larmor frequencies in the horizontal direction throughout the time it is applied. It thus creates proton columns, which all have an identical Larmor frequency.
As this gradient is applied simultaneously on receiving the signal, the frequency data is included.
Interpreting spatial encoding in MRI
Slice selection
Selecting a slice plane and spatially encoding each voxel involves the use of magnetic field gradients. Magnetic field intensity varies regularly along the gradient application axis. Each gradient is characterized by its strength (greater or lesser field variation for the same unit of distance), direction and the moment and time of application. The slice selection gradient modifies the precession frequency of the protons such that an RF pulse with the same frequency will cause them to shift (resonance). The slice selection gradient is simultaneously applied to all the RF pulses. Through the intermediary of the slice selection gradient, RF pulse frequency corresponds to selectivity in space. RF pulse bandwith and waveform determine slice thickness and profile.
Phase encoding
Each phase encoding step acts as a kind of sieve letting through regularly spaced horizontal signals. This filter is sensitive to the vertical spatial distribution of the signals in the slice plane. The greater the phase difference, the thinner and finer the filter. In the absence of phase encoding, the signal will come from the whole slice. This is why multiple phase encoding steps are needed to acquire enough data to reconstruct the image. By analyzing the signals obtained with hundreds of different profiles, corresponding to as many fine combs, an image can be reconstructed (not only composed of horizontal bands, but of more complex contours).
To carry out the different phase encoding steps, the gradient is applied with different, regularly incremented values. It is bipolar, that is to say gradients are used with positive and negative values that are symmetrical to 0. In terms of the « filter size » range, a bipolar gradient is equivalent to a gradient that is only positive, for instance, and of the same absolute variation amplitude. However this requires positive amplitudes that are twice as high (causing greater dephasing and a poorer quality signal).
Frequency encoding
Frequency encoding can be interpreted in the same way, in the horizontal direction. When the frequency-encoding gradient is applied, the signal is digitized at regular intervals in time. Each signal sample corresponds to a given accumulation of the gradient effect on the whole splice signal: the longer the time, the longer the effect of the gradient on the spins, and the greater their phase modification. We see the same filter effect, sensitive to spatial distribution in the horizontal direction (the direction in which the gradient is applied). To obtain the equivalent of a bipolar effect, a frequency encoding gradient half-lobe is applied but in the opposite direction (dephasing lobe) prior to signal recording.
All the signals from the same slice are recorded in k-space then processed to form an image of the slice plane.
While frequency spatial encoding only takes a few milliseconds of signal reading, phase spatial encoding involves repeating the imaging sequence. In a classic spin echo sequence, a single phase encoding step is performed during each repetition time (TR). As TR values can be of up to 3 seconds, phase encoding is therefore much longer than frequency encoding.
3D spatial encoding
The particularities of 3D acquisitions are:- excitation of a complete volume at each repetition (volume = « thick slice »), rather than one thin slice
- spatial encoding in 3D by adding phase encoding in the 3rd dimension in relation to the phase and frequency encodings used in 2D imaging
- multiplication of the number of repetitions of a factor equal to the number of « slices» (partitions) in the third dimension to fill all the 3D k-space
- reconstruction by 3D Fourier transform.
All this has consequences for:
- acquisition time: given the large amount of acquired data needed to fill the 3D k-space, either very short TR sequences (echo gradient type) are used, or faster k-space filling methods.
- the amount of signal: at each repetition, the signal comes from the whole volume, rather than a single slice. Therefore more signal is recorded, with fewer noise. The partitions can be finer than the classic 2D slices, because the signal to noise ratio is better compared to a slice of the same thickness acquired in 2D.
- spatial resolution: the entire volume of interest is explored, with no interval between partitions, allowing fine section and multiplanar rconstructions.
- artifacts: because of the two phase encodings, wraparound and truncation artifacts can be seen in two different directions
Duration of a 3D imaging sequence
Duration = TR ∙ NPy ∙ NPz ∙ NexWith
- TR = Repetition time
- NPy = Number of encoding steps in the y-axis
- NPz= Number of encoding steps in the z-axis
- Nex= Number of excitations
MRI image formation
Summary
The Physics behind it all
- Introduction
- Fourier transform
- Spatial frequency
- 2D Fourier transform
- K-space
- Gradients and spatial frequency
- Spatial frequency, image contrast and resolution
- K-space linear filling
Learning objectives
After reading this chapter, you should be able:- To explain the relation between time domain, frequency domain and Fourier transform
- To define: spatial frequency, phase and magnitude
- To draw the effects of a point in k-space on the image
- To state the relations between RF pulses, gradients and navigation in k-space
- To describe the k-space trajectory with a spin echo sequence
- To link contrast, spatial resolution and field of view to k-space
Key points
- Spatial encoding in MR imaging uses magnetic field gradients. These gradients allow the encoding of spatial data as spatial frequency information. These data are mapped into k-space so that an inverse 2D Fourier transform reconstructs the MR image.
- The location of the data in k-space depends on the strength and the duration of the gradients. If no gradient is applied (or if the net effect of the gradient is null), the location is at the center of k-space. The higher is the net strength of the gradient, or the longer the gradient is applied, the farther from the origin of k-space the data will be located. Gradients are bipolar (negative or positive), so it is possible to move in opposite directions from the center (to the left or the right, and to the top or the bottom).
- Each point of k-space encodes for spatial information of the entire MR image.
- Each point of the MR image is the result of the combination of all the data of k-space.
- The data along a line passing through the center of k-space represent the Fourier transform of the projection of the image onto a line with the same orientation. In other words, a point in k-space encodes for image signal variations in the same direction as the line passing through this point and the center of k-space.
- Center of k-space contains low-spatial-frequency information.
- Most image information is contained in low-spatial-frequency information, corresponding to general shape and contrast.
- Periphery of k-space does not correspond to periphery of the image : it contains high-spatial-frequency information. The higher the spatial frequency, the smaller are the details of the image (edges and spatial resolution).
References
|
Introduction
The readout MR signal is a mix of RF waves with different amplitudes, frequencies and phases, containing spatial information. This signal is digitized and raw data are written into a data matrix called K-space. K-space data are equivalent to a Fourier plane. To go from a k-space data to an image requires using a 2D inverse Fourier Transform.This chapter introduces the most complex concepts of MRI image formation for many students, so take to your time to follow this step.
Understanding k-space and Fourier transform will help you understand the trade offs that are made for fast imaging and sequences optimisation, as well as the source of various artifacts.
Fourier transform
The 1D Fourier transform is a mathematical procedure that allows a signal to be decomposed into its frequency components.On the left side, the sine wave shows a time varying signal.
On the right side, you can observe its equivalent in the frequency domain. The sine wave corresponds to a plot at the same frequency as the sine wave and same amplitude as the maximal amplitude of the sine wave.
Single frequency
The following animation demonstrates the relation between time domain and frequency domain about a sine wave representing a single frequency sound.Moving scroll bars change frequency and amplitude of the sound so that you can see modifications in time and frequency domain.
To describe a sine wave, we need its amplitude and frequency, but also its phase. The following animation illustrates the consequence of a phase change on the sine wave (shift in time domain)
Multiple frequencies
The Fourier transform is a mathematical procedure that decomposes a signal into a sum of sine waves of different frequencies, phases and amplitude. The human ear does the same processing with sounds, which are analyzed as a spectrum of elementary frequencies. Knowing frequency, amplitude and phase of each sine wave, it is possible to reconstruct the signal (inverse Fourier transform).By moving the scroll bars, you can manipulate up to 3 sine waves of different frequencies, amplitudes and phases.
On the left side, you can see the graph of their sum.
Sounds
Even if a signal is very complex, the Fourier transform will always be able to decompose it into its frequency components from which we will reconstruct a signal very similar to the original. (Representing a signal exactly would require an infinite number of frequency components, which is not possible in practice)Even music can be decomposed by a Fourier Transform. This is employed in spectrum analyser (left side).
Note that low frequency components have the highest amplitude.
Spatial frequency
Fourier transform is able to decompose images. Instead of analyzing a time varying signal, it decomposes a variation of intensity (gray levels) over distance. Time domain becomes space domain (time variable replaced by X-coordinate) and frequency is called spatial frequency.
As frequency refers to the (inverse of the) periodicity with which the sound sine wave repeats, the spatial frequency refers to the (inverse of the) periodicity with which the image intensity values change.
Image features that change in intensity over short image distances correspond to high spatial frequencies.
Image features that change in intensity over long image distances correspond to low spatial frequencies.
2D Fourier transform and MRI image reconstruction
To decompose a 2D image, we need to perform a 2D Fourier transform. The first step consists in performing a 1D Fourier transform in one direction (for example in the row direction Ox). In the following example, we can see:- the original image that will be decomposed row by row
- the gray level intensities of the choosen line
- the spectrum obtained after 1D Fourier transform
The second step of 2D Fourier transform is a second 1D Fourier transform in the orthogonal direction (column by column, Oy), performed on the result of the first one.
The final result is called Fourier plane that can be represented by an image.
In this example, here is how to read the Fourier plane:
- Horizontal and vertical axis correspond to horizontal and vertical spatial frequencies
- Pixel intensity corresponds to the amplitude (or magnitude) of frequency component
- Color corresponds to the phase of frequency component.
K-space exploration
The readout MR signal is stored in K-space which is equivalent to a Fourier plane. To go from a k-space data to an image requires using a 2D inverse Fourier Transform.
Frequency-encoding and phase-encoding are done so that data is spatially encoded by differences in frequency and phase, amenable to analysis by Fourier transform. In k-space, fx-coordinates (horizontal spatial frequencies) and fy-coordinates (vertical spatial frequencies) of the Fourier plane are replaced by kx and ky-coordinates.
A classic spin echo sequence fills the k-space line by line. Here is the explanation of the k-space trajectory :
- 90° RF pulse + Slice-selection gradient : location at origin (center) of k-space
- Negative and strong phase-encoding gradient: moves to the lower bound of k-space
- Positive frequency-encoding gradient (dephasing lobe): moves to the right bound, location at the lower right corner
- 180° RF pulse + Slice-selection gradient : moves to the opposite location, location at the upper left corner
- Positive frequency-encoding gradient + Data acquisition: moves to the right + acquire MR signal
- Repetition for each line with increasing phase-encoding gradient strength (negative to positive intensity). The amount of gradient phase change between adjacent line is constant. This results in a sequential (line by line) filling of k-space from top to bottom.
- The initial RF excitation pulse (with the slice-selection gradient) is the beginning of the sequence: location is at the center of k-space.
- A 180° RF pulse causes a jump to the opposite location.
- The greater the net strength of the phase-encoding gradient (or the longer the gradient is on), the farther from the k-space origin the data belong, in the upper direction if the gradient is positive or in the lower direction if the gradient is negative. As the duration of phase-encoding gradient is most often constant, the strength of the phase-encoding gradient governs the location on the vertical axis (ky-coordinate).
- The longer the frequency-encoding gradient is on (or the greater the net strength of the gradient is), the farther from the k-space origin the data belong, in the right direction if the gradient is positive or in the left direction if the gradient is negative. As the strength of the frequency-encoding gradient is most often constant, the duration of the frequency-encoding gradient governs the location on the horizontal axis (kx-coordinate)
Gradients and spatial frequency
How and when the MR signals are mapped into the k-space cause great differences in the spatial, temporal and contrast resolution of the resulting MR images.The location of the datas in k-space depends on the net strength and duration of the phase encoding gradient and frequency encoding gradient:
- A low-amplitude or short-duration gradient event encodes low-spatial-frequency information
- A high-amplitude or long-duration gradient event encodes high-spatial-frequency information
Drag the cross to see the image corresponding to low or high-spatial frequency information. A point in k-space encodes for data with the same orientation of the line passing through this point and the center of k-space.
Spatial frequency, image resolution and contrast
Most MR image information (contrast and general shape) is contained in the center of k-space. Low-spatial-frequency data have the highest amplitude, giving the greatest changes in gray levels (contrast). However, these changes spread over in the image and only give the general shape of organs.We can see below the resulting images of inverse 2D transform performed on data at the center of k-space. Image is contrasted but blurry.
High-spatial-frequency data have a lower amplitude. They don't have effect on contrast or general shape but sharpens the image as they encode the edges (rapid changes of image signal as a function of position). Thus, the farther from the center of k-space the data are collected, the higher is the spatial-frequency information and the better the spatial resolution will be.
We can see below the resulting images of inverse 2D transform performed on data at the periphery of k-space. As they have a low intensity, click on Enhance contrast to see the effect of high-spatial-frequency information on image.
In order to make visible the effect of low and high spatial frequencies, you can choose in the following experiment the frequencies you want to keep to reconstruct the image. The low spatial frequencies are at the center of k-space while the high spatial frequencies are at its periphery.
Then, move the blue borders to change the extent of k-space data to be used : this allows for selecting the lowest or highest spatial frequencies.
If the image seems too dark (particularly if you select high spatial frequencies), click the Amplify button to enhance image contrast.
Linear trajectory through k-space
The easier way to fill the k-space is to use a
line-by-line rectilinear trajectory. One line of k-space is fully acquired at
each excitation, containing low and high-horizontal-spatial-frequency
information (contrast and resolution in the horizontal direction). Between each
repetition, there is a change in phase-encoding-gradient strength,
corresponding to a change in ky-coordinate. This allows filling of all the
lines of k-space from top to botton, acquiring high-positive
vertical-spatial-frequency information (resolution in the vertical direction)
then low vertical-spatial-frequency information (contrast in the vertical
direction) and then high-negative vertical-spatial-frequency information
(resolution in the vertical direction again).
During the filling of k-space, the resulting image is containing at the bigenning the edge information with low contrast, then the general shape and contrast with a blur in the vertical direction that will disappear as high-vertical-spatial-frequency information are completed.
During the filling of k-space, the resulting image is containing at the bigenning the edge information with low contrast, then the general shape and contrast with a blur in the vertical direction that will disappear as high-vertical-spatial-frequency information are completed.
Sequences
Summary
MRI Sequences
- Introduction
- Characteristics
- Classification
- Acronyms
- Spin echo
- Fast spin echo
- Ultrafast spin echo
- Inversion Recovery / STIR / FLAIR
- Gradient echo
- Spoiled gradient echo
- Ultrafast spoiled gradient echo
- Steady-state gradient echo
- T2-enhanced steady-state gradient echo
- Balanced gradient echo
- Echo planar imaging (EPI)
- Hybrid echo (spin echo + gradient echo)
Learning objectives
After reading this chapter, you should be able:- To present the members of the spin echo and gradient echo sequence families
- Describe the principles of signal acquisition for each type of sequence
- Explain the contrast obtained, the advantages and disadvantages of each sequence
- Present the sequence acceleration techniques in spin echo
- Examine the effect of inversion-recovery on contrast and its applications
- Define the relationship between TR, flip angle and longitudinal magnetization in gradient echo
- Describe the notion of steady state transverse magnetization in gradient echo, the conditions in which it occurs and its impact on the sequences
- Explain the methods of echoplanar imaging and compensations for the gain in speed
Key points
Type of sequence
|
Principles
|
Advantages
|
Disadvantages
|
Spin echo (SE)
|
simple, SE
T1, T2, DP contrast |
Contrast
|
Slow (especially in T2)
|
Multiecho SE
|
SE several TE, several images
|
DP + T2 images
|
Slow, even if acquisition of the 2nd image does not
lengthen acquisition
|
Fast SE
|
SE, echo train
effctive TE |
Faster than simple SE simple
ES contrast |
Fat shown as a hypersignal
|
Ultrafast SE
|
SE, long echo train, half-Fourier
|
Even faster
|
Low signal to noise ratio
|
IR
|
RF 180°, TI + ES/ESR/EG
|
T1 weighting
Tissue suppression signal if TI is adapted to T1 |
Longer TR / acquisition time
|
STIR
|
IR, short TI 150 ms
|
Fat signal suppression
|
Longer TR / acquisition time
|
FLAIR
|
IR, long TI 2200 ms
|
CSF signal suppression
|
Longer TR / acquisition time
|
Gradient echo (GE)
|
< 90° α and short TR
No rephasing pulse |
+ speed
|
T2* not T2
|
GE with spoiled residual transverse magnetization
|
TR < T2
Gradients / RF dephasers |
T1, DP weighting
|
|
Ultrafast GE
|
small α and very short TR
Gradients / RF dephasers k-space optimization |
++ speed
cardiac perfusion |
Poor T1 weighting
|
Ultrafast GE with magnetization preparation
|
+ preparation pulse:
- IR (T1weighted) - T2 sensibilization |
++ speed
AngioMRI Gado Cardiac perfusion / viability |
|
Steady state GE
|
TR < T2
Rephasing gradients FID |
+ signal
++ speed |
Complex contrast
|
Contrast enhanced steady state GE
|
Rephasing gradients
Hahn echo ( trueT2) |
Not much signal
T2 weighted |
|
Balanced
steady state GE |
Balanced gradients in all 3 directions
T2/T1contrast |
++ signal, ++ speed
Flow correction |
|
Echoplanar
|
Single GE or multi shot
Preparation by SE (T2), GE (T2*), IR (T1), DW Exacting for gradients |
++++ speed
Perfusion MRIf BOLD Diffusion |
Limited resolution
Artifacts |
Hybrid echo
|
Fast SE
+ intermediary GE |
++ speed
SAR reduction |
References
|
Sequences
MRI is the imaging technique that has most benefited from technological innovation. The many advances have led to improvements in quality and acquisition speed.Each sequence is a subtle combination of radiofrequency pulses and gradients. Whatever the type of sequence, the aims are to favor the signal of a particular tissue (contrast), as quickly as possible (speed), while limiting the artifacts and without altering the signal to noise ratio. There are over a hundred different sequences and to complicate things further, manufacturers tend to each choose their own acronyms!
Characteristics of an MRI sequence
The architecture of a sequence consists of the essential components on one hand, and the various options, on the other. The building blocks of the sequence are radiofrequency pulses and gradients.The essential components for any imaging sequence are:
- An RF excitation pulse, required for the phenomenon of magnetic resonance
- Gradients for spatial encoding (2D or 3D), whose arrangement will determine how the k-space is filled
- A signal reading, combining one or a number of echo types (spin echo, gradient echo, hahn echo, stimulated echo…) determining the type of contrast (the varying influence of relaxation times T1, T2and T2*).
- Either modify the contrast (preparing magnetization by inversion-recovery, fat saturation, magnetization transfer…)
- or accelerate the sequence (partial Fourier plane filling, parallel acquisition, fast magnetization restoration…)
- or to reduce artifacts (flow compensation, synchronisation, presaturation bands …)
With the exception of inversion-recovery, the optional techniques to modify the contrast of a sequence will be examined in a dedicated chapter. Likewise, the treatment of artifacts and parallel imaging methods will be dealt with in separate chapters.
Sequence classification
There are two main sequence families, depending on the type of echo recorded:
- spin echo sequences, characterized by the presence of a 180° rephasing RF pulse
- gradient echo sequences
- Fast spin echo sequences, Single shot FSE and Haste
- Gradient echo sequences with spoiling of residual transverse magnetization (spoiled gradient echo and ultrafast gradient echo), a group of gradient echo sequences with steady state residual transverse magnetization (Steady state gradient echo) and its derivatives (Contrast enhanced steady state gradient echo) and with balanced gradients (Balanced steady state gradient echo), echoplanar (EPI).
Magnetic resonance angiography sequences (FBI, contrast-enhanced MRA, TOF, PC) perfusion imaging, diffusion imaging (DW) and MR spectroscopy will be dealt with in separate chapters in the second part.
Sequences acronyms
Due to manufacturers each using their own terminology to denominate their sequences, there are no standard denominations for each common type of sequence.Here is a table of the equivalent manufacturers’ acronyms with the corresponding type of sequence.
Type of sequence
|
Philips
|
Siemens
|
GE
|
Hitachi
|
Toshiba
|
Spin Echo (SE)
|
SE
|
SE
|
SE
|
SE
|
SE
|
Multi echo SE
|
Multi SE
|
Multi écho
MS |
SE
|
SE
|
Multi écho
|
Fast SE
|
Turbo SE
|
Turbo SE
|
Fast SE
|
Fast SE
|
Fast SE
|
Ultra fast SE
|
SSH-TSE
UFSE |
SSTSE
HASTE |
SS-FSE
|
FSE - ADA
|
(Super)FASE
DIET |
IR
|
IR
IR TSE |
IR/IRM
TurboIR/TIRM |
IR
FSE-IR |
IR
FIR |
IR
Fast IR |
STIR
|
STIR
STIR TSE |
STIR
Turbo STIR |
STIR
Fast STIR |
STIR
Fast STIR |
STIR
Fast STIR |
FLAIR
|
FLAIR
FLAIR TSE |
FLAIR
Turbo FLAIR |
FLAIR
Fast FLAIR |
FLAIR
Fast FLAIR |
FLAIR
Fast FLAIR |
Gradient echo (GE)
|
FFE
|
GRE
|
GRE
|
GE
|
FE
|
Spoiled GE
|
T1-FFE
|
FLASH
|
SPGR
MPSPGR |
RSSG
|
RF-spoiled FE
|
Ultra fast GE
|
T1-TFE T2-TFE THRIVE |
TurboFLASH VIBE |
FGRE
Fast SPGR FMPSPGR VIBRANT FAME LAVA |
SARGE
|
Fast FE RADIANCE QUICK 3D |
Ultrafast GE with magnetization preparation
|
IR-TFE
|
T1/T2-TurboFLASH
|
IR-FSPGR
DE-FSPGR |
Fast FE
|
|
Steady state GE
|
FFE
|
FISP
|
MPGR, GRE
|
TRSG
|
FE
|
Contrast enhanced steady state GE
|
T2-FFE T2
|
PSIF
|
SSFP
|
FE
|
|
Balanced GE
|
Balanced
FFE |
True FISP
|
FIESTA
|
BASG
|
True SSFP
|
SE-Echo planar
|
SE-EPI
|
EPI SE
|
SE EPI
|
SE EPI
|
SE EPI
|
GE-Echo planar
|
FFE-EPI
TFE-EPI |
EPI Perf
EPIFI |
GRE EPI
|
SG-EPI
|
FE-EPI
|
Hybrid echo
|
GRASE
|
TGSE
|
Hybrid EPI
|
Spin echo
Type of sequence
|
Philips
|
Siemens
|
GE
|
Hitachi
|
Toshiba
|
Spin Echo (SE)
|
SE
|
SE
|
SE
|
SE
|
SE
|
Generic diagram
Gradients and phase in spin echo sequences
The rephasing lobe of the slice selection gradient, the phase encoding gradient and the dephasing lobe of the readout gradient are applied simultaneously, immediately after the excitation pulse.The slice selection gradient applied for the 180° pulse requires no rephasing lobe. However, two identical gradient lobes are applied on either side of this gradient to eliminate the transverse magnetization created by the 180° rephasing pulse on the edge of the slice (where the protons will in fact be submitted to a flip angle of less than 180° due to the imperfect slice profile).
Duration of a spin echo sequence
Duration = TR ∙ NPy ∙ NexWith
- TR = Repetition time
- NPy = Number of phase encoding steps
- Nex = Number of excitations
Contrast
A spin echo sequence has two essential parameters: TR and TE.TR is the time interval between two successive 90° RF waves. It conditions the longitudinal relaxation of the explored tissues (depending on T1). The longer the TR, the more complete the longitudinal magnetization regrowth (Mz tends to M0). Reducing TR will weight the image in T1 as the differences between the longitudinal relaxation of the tissues’ magnetization will be highlighted . In classic spin echo, after TR time, a single k-space line will be acquired. TR repetition is thus responsible for the duration of the sequence.
TE is the time interval between the 90° flip and receipt of the echo, the signal being produced by transverse magnetization. Transverse magnetization decreases according to the time constant T2 of each tissue (the field heterogeneities being compensated by the 180° flip applied at TE/2).
In the T2–weighted spin echo sequence the TR and TE parameters are optimized to reflect T2 relaxation.
When the TR is long (over 2000 milliseconds), longitudinal magnetization recovery is complete and on the following flip, the influence of T1 on signal magnitude will be minimized. Associated with long TE (80 to 140 milliseconds), the different tissues are better highlighted according to their T2.
Long T2 tissues will appear as a hypersignal, as opposed to short T2 structures, which will appear as a hyposignal.
The proton density weighted spin echo sequence has optimized TR and TE parameters to minimize the influence of both T2 and T1. The contrast obtained will depend on the density of the hydrogen nuclei (i.e. protons).
A long TR (over 2000 milliseconds), associated with a short TE (10 to 20 milliseconds) will relatively suppress both the influence of T1 and the effect of T2 on signal magnitude.
Interest
Historically, spin echo was the first sequence to be used. It has been a benchmark for all subsequent developments, namely in terms of contrast. The 180° rephasing pulse gives a « true T2 » signal rather than a T2*signal.Choosing the right sequence parameters (TR and TE) will produce images weighted in T1, T2 or proton density. The major disadvantage with T2 weighted spin echo sequences is linked to long TR resulting in prohibitive acquisition times.
While spin echo sequences can be used in clinical practice to obtain good quality anatomical T1-weighted images, faster types of sequence are preferred to obtain T2-weighted images.
Fast spin echo
Type of sequence
|
Philips
|
Siemens
|
GE
|
Hitachi
|
Toshiba
|
Multi-echo SE
|
Multi SE
|
Multi écho
MS |
Multi écho
|
||
Fast SE
|
Turbo SE
|
Turbo SE
|
Fast SE
|
Fast SE
|
Fast SE
|
Fast spin echo
In fast spin echo sequences, the interval of time after the first echo, is used to receive the echo train, to fill the other k-space lines in the same slice . Because of the reduced number of repetitions (TR) required, the k-space is filled faster and slice acquisition time is reduced.This is done by applying new 180° pulses to obtain a spin echo train. After each echo, the phase-encoding is cancelled and a different phase-encoding is applied to the following echo.
The number of echoes received in the same repetition (during TR time) is called the Turbo Factor or Echo Train Length (ETL).
Multi-echo SE sequences
These sequences allow several images of the same slice position without increasing overall acquisition time. The advantage is that the images are obtained with a different contrast, which is useful in characterizing certain lesions (for example, highlighting contrast at long TE for hepatic angioma, which appears as a relative hypersignal).
After the first echo is obtained, there is a free interval until the next TR. By applying a new 180° pulse, a new echo is received, with the same phase encoding, to build the second image . The echo time of the 2 images differs and the second image will be more T2 weighted than the first.
Typically, these sequences are used to obtain simultaneously PD- and T2-weighted images.
Contrast, resolution, scan time
The contrast in fast spin echo is modified in relation to a standard spin echo sequence. As the echoes are received at different echo times, the echoes corresponding to the central k-space lines are the ones that will determine image contrast. The moment at which theses echoes are acquired is called effective TE.In T1 weighted sequences, the need to choose a short TR limits echo train length. This type of sequence is very commonly used in T2 weighting, namely in pelvic imagery.
Fat signal in fast spin echo
Within lipid molecules a spin-spin coupling (J coupling) occurs between the atomic nuclei. This coupling shortens relaxation time T2. Fast repetition of 180° pulses in fast spin echo sequences will perturb J coupling, causing fat T2 to lengthen.Thus, fat has a higher T2 signal in fast spin echo than in standard spin echo, the latter respecting J coupling.
A DIET (Delayed Interval Echo Train) sequence is a fast SE sequence where the delays between 180° pulses are designed to respect J coupling: as a result, the fat maintains an appearance closer to that observed in a standard SE sequence.
Interest and limits
The interest of fast SE sequences resides in their speed (around ten seconds) added to their low sensitivity to magnetic susceptibility artifacts and magnetic field heterogeneities.Modifications in contrast and fat signal must be taken into account in interpreting the images. The risk of artifacts and the large quantity of radiofrequency energy deposited by 180° pulses restricts the parameters (TR, effective TE, echo train length) of this type of sequence.
Fast spin echo can be combined with the technique developed for multi-echo sequences to obtain images faster with different contrasts in the same zone of interest.
Ultrafast spin echo sequences
Type of sequence
|
Philips
|
Siemens
|
GE
|
Hitachi
|
Toshiba
|
Ultrafast SE
|
SSH-TSE
UFSE |
SSTSE
HASTE |
SS-FSE
|
FSE - ADA
|
(Super)FASE
DIET |
Generic diagram
The echo train technique can be pushed to the limit to fill the entire Fourier plane with a single 90° pulse (TR is thus infinite) . These so-called « single-shot » sequences require the successive application of as many 180° pulses as there are k-space lines to fill.
The sequence can be further accelerated, avoiding the need to register the latest echoes (whose signal is much reduced by T2 relaxation) by partial k-space acquisition. Just over half the k-space lines are actually acquired and the missing lines are calculated using k-space symmetry properties. This reduces acquisition time by a factor close to 2, but to the detriment of the signal to noise ratio of the image.
Contrast and scan time
Given the length of the echo train, the images obtained are highly T2 weighted, since the majority of k-space lines are filled with long TE echoes.With this type of sequences, a slice can be made in under a second.
Duration of an ultrafast spin echo sequence
Duration = TE • number of phase encodings to acquireInterest and limits
These sequences are well adapted to imaging non-circulating liquid structures appearing as a T2-weighted hypersignal (Cholangio-MRI and Uro-MRI) . Due to their rapidity, they have low sensitivity to movement and are compatible with apnea (mobile structures: liver, abdomen, heart).The negative impact of very long echo trains is a decay in signal to noise ratio (weak signal from late echoes and very high effective TE) with low spatial resolution and blurring in the phase encoding direction.
Inversion Recovery, STIR and FLAIR
Type of sequence
|
Philips
|
Siemens
|
GE
|
Hitachi
|
Toshiba
|
IR
|
IR
IR TSE |
IR/IRM
TurboIR/TIRM |
IR
FSE-IR |
IR
FIR |
IR
Fast IR |
STIR
|
STIR
STIR TSE |
STIR
Turbo STIR |
STIR
Fast STIR |
STIR
Fast STIR |
STIR
Fast STIR |
FLAIR
|
FLAIR
FLAIR TSE |
FLAIR
Turbo FLAIR |
FLAIR
Fast FLAIR |
FLAIR
Fast FLAIR |
FLAIR
Fast FLAIR |
Generic diagram
Inversion-recovery is a magnetization preparation technique followed by an imaging sequence of the spin echo type in its « standard » version .
The sequence starts with a 180° RF inversion wave which flips longitudinal magnetization Mz in the opposite direction (negative). Due to longitudinal relaxation, longitudinal magnetization will increase to return to its initial value, passing through null value.
To measure the signal, a 90° RF wave is applied to obtain transverse magnetization. The delay between the 180° RF inversion wave and the 90° RF excitation wave is referred to as the inversion time TI.
As longitudinal regrowth speed is characterized by relaxation time T1, these sequences are weighted in T1. Inversion-recovery also increases weighting of the associated imaging sequence (spin echo or gradient echo of varying speeds).
With this type of sequence, certain tissues have a negative signal. In terms of display, two possibilities exist :
- Either signal magnitude (amplitude in relation to 0) used for gray scale display: the more absolute the value of the tissue signal (positive or négative), the stronger it will be.
- Or the gray levels will be distributed from the negative signal values to the positive values (with a null signal background that will be gray rather than black): this is the « true » display type.
Inversion-recovery can be combined with sequence types other than the standard spin echo. In particular, it can be used with fast spin echo sequences, to save considerable time, as inversion-recovery requires relatively long TR to allow magnetization the time to regrow. Iinversion-recovery also serves as magnetization preparation for gradient echo sequences, to weight them in T1.
STIR sequences
The advantage of these sequences is that they offer a fat signal suppression technique with low sensitivity to magnetic field heterogeneities or to the effects of magnetic susceptibility in the presence of metal (orthopedic prostheses in osteoarticular imaging for example). They can be used with T1 or T2 weighting (particularly in spin echo sequences where fat appears as a hypersignal).
This technique must not be used to suppress a fat signal gadolinium injection because gadolinium–enhanced tissues have a shortened T1 and may be erased by short TI inversion-recovery (which is not specific to tissue but to its relaxation time T1).
FLAIR sequences
These sequences are routinely used in cerebral MRI for edema imaging.
Gradient echo
Type of sequence
|
Philips
|
Siemens
|
GE
|
Hitachi
|
Toshiba
|
Gradient echo (GE)
|
FFE
|
GRE
|
GRE
|
GE
|
FE
|
Characteristics
The gradient echo sequence differs from the spin echo sequence in regard to:- the flip angle usually below 90°
- the absence of a 180° RF rephasing pulse
The advantages of low-flip angle excitations and gradient echo techniques are faster acquisitions, new contrasts between tissues and a stronger MR signal in case of short TR.
The flip angle determines the fraction of magnetization tipped in the transverse plane (which will produce the NMR signal) and the quantity of magnetization left on the longitudinal axis.
If the flip angle decreases, the residual longitudinal magnetization will be higher and the recovery of magnetization for a given T1 and TR will be more complete. On the other hand, the result of a lower flip angle excitation is a lower tipped magnetization.
The actual decay of the transverse magnetization is due to several factors:
- spin-spin tissue-specific relaxation (T2) which is random
- B0 field inhomogeneities and magnetic susceptibility, which are static
Gradient echo
As there is no 180° RF pulse, a bipolar readout gradient (which is the same as the frequency-encoding gradient) is required to create an echo. The gradient echo formation results from applying a dephasing gradient before the frequency-encoding or readout gradient.The goal of this dephasing gradient is to obtain an echo when the readout gradient is applied and the data are acquired. The dephasing stage of the readout gradient is in the inverse sign of the readout gradient during data acquisition. Moreover, its dephasing effect is designed so that it corresponds to half of the dephasing effect of the readout gradient during data acquisition. Consequently, during data acquisition, the readout gradient will rephase the spins in the first half of the readout (by reversing the dephasing effect of the dephasing lobe), and the spins will dephase in the second half (due to the dephasing effect of the readout gradient). The time during which the peak signal is obtained is called Echo Time (TE).
Steady state
In gradient echo, TR reduction may cause permanent residual transverse magnetization in TR below T2: the transverse magnetization will not have completely disappeared at the onset of the following repetition and will also be submitted to the flip caused by the excitation pulse.Two main classes of gradient echo sequence can be distinguished, depending on how residual transverse magnetiztion is managed:
- gradient echo sequences with spoiled residual transverse magnetization
- steady state gradient echo sequences that conserve residual transverse magnetization and therefore participate in the signal.
Spoiled gradient echo sequences
Type of sequence
|
Philips
|
Siemens
|
GE
|
Hitachi
|
Toshiba
|
Spoiled GE
|
T1-FFE
|
FLASH
|
SPGR
MPSPGR |
RSSG
|
RF-spoiled FE
|
Principles
In certain cases, the steady state can be detrimental, namely for obtaining T1 weighted sequences. To resolve this problem, gradients and/or RF pulses (spoilers) are used to eliminate residual transverse magnetization.In this type of echo gradient sequence, image weighting will depend on:
- the flip angle for T1 weighting (the greater the angle, the more T1 weighting)
- the TE for T2* weighting (the shorter the TE, the more T2* deweighting)
These sequences allows for fast 3D imaging during short apnea (10 to 20 seconds). They are used in angiography after gadolinium injection, possibly completed by a prior acquisition to allow subtraction. This type of sequence is also used in intra-voxel water-fat mixture imaging, by choosing in phase and out of phase TEs (cf. Chemical shift artifact of the second type). Like all other gradient echo sequences, these are also sensitive to magnetic susceptibility artifacts.
Advanced topics
Spoiling gradients and RF pulses
The spoiing RF pulse consists in randomly varying the excitation pulse phase at each repetition.A spoiling gradient has 2 components:
- A constant component, consisting of prolonging the readout gradient
- A variable component, changing randomly at each repetition, in the direction of the slice selection gradien
Ernst angle and TR in gradient echo sequences
For a given flip angle and relaxation time T1, there is an optimal TR to obtain the maximum signal after a series of excitation pulse . The relationship between flip angle and optimal TR follows the Ernst angle formula:
Ultrafast spoiled gradient echo sequences
Type of sequence
|
Philips
|
Siemens
|
GE
|
Hitachi
|
Toshiba
|
Ultrafast spoiled GE
|
T1-TFE T2-TFE THRIVE |
TurboFLASH VIBE |
FGRE
Fast SPGR FMPSPGR VIBRANT FAME LAVA |
SARGE
|
Fast FE RADIANCE QUICK 3D |
Ultrafast spoiled GE
with magnetization preparation |
IR-TFE
|
T1/T2-TurboFLASH
|
IR-FSPGR
DE-FSPGR |
Fast FE
|
Ultrafast gradient echo sequences use a small flip angle, a very short TR and optimized k-space filling to reduce acquisition time (to roughly one second per slice). The drawback of a small flip angle and very short TR is poor T1-weighting.
To preserve T1 contrast, a 180° inversion pulse prepares magnetization before repetitions of the ultrafast gradient echo imagery sequence . Effective inversion time will correspond to the delay between the inversion pulse and acquisition of the central k-space lines.
To obtain T2-weighting, the preparatory pattern is of the spin echo type (90° - 180°) so that the ultrafast GE imaging sequence can start with longitudinal magnetization whose amplitude depends on T2 .
K-space filling can follow variable trajectories (linear, centripetal, centrifugal) to determine image contrast. Either all the k-space lines are acquired after a single inversion pulse (« single shot »), or only a group of lines is acquired (segmented filling).
The speed of these T1-weighted gradient echo sequences is particularly suited to monitoring Gadolinium bolus arrival in imaging at arterial phase and in T1 high resolution 3D imaging.
Gradient echo sequences with steady state residual transverse magnetization
In steady state gradient echo sequences, residual transverse magnetization is conserved. This will participate in the signal and the contrast and vary according to the type of sequence.By maintaining residual transverse magnetization, excitation pulses will produce new echos (Hahn echos, stimulated echos) in addition to the gradient echo that depends on the free precession signal (FID).
There are several variants in the family of steady state gradient echo sequences, according to the type of echo recorded (which determines contrast) and how the gradients are adjusted.
Steady-state gradient echo
Type of sequence
|
Philips
|
Siemens
|
GE
|
Hitachi
|
Toshiba
|
Steady state GE
|
FFE
|
FISP
|
MPGR, GRE
|
TRSG
|
FE
|
Steady-state gradient echo
In « standard » steady state gradient echo sequences:
- phase encoding is cancelled at the end of each repetition
- only the echo corresponding to the free induction signal (FID) is recorded
For short TR (less than T2) and fairly large flip angles (40° – 90°), the contrast of this type of sequence varies according to T2/T1 ratio.
Hahn echo
Two RF pulses at the same excitation angle produce a Hahn echo (partial spin echo), whose amplitude depends on T2. With two 90° pulses, we obtain:- Magnetiztion flip after the first 90°RF excitation pulse
- Dephased transverse magnetization
- Magnetization flip after the second 90° RF excitation pulse
- Only the remaining transverse component is taken into account for the Hahn echo
- Partial rephasing occurs
- in the le transverse plane
- Maximum rephasing corresponds to the peak of the Hahn echo
- Dephasing then continues in the transverse plane
Magnetization
Transverse magnetization
T2-enhanced steady-state gradient echo
Type of sequence
|
Philips
|
Siemens
|
GE
|
Hitachi
|
Toshiba
|
T2-enhanced steady-state GE
|
T2-FFE T2
|
PSIF
|
SSFP
|
FE
|
In T2-enhanced steady-state gradient echo sequences:
- residual transverse magnetization is conserved
- the sequence is inverted in time, compared to the preceding sequences
- only the echo corresponding to the Hahn echo, dependent on T2 but weaker tthan spin echo, is recorded
The weak intensity of the Hahn echo, produces a low signal to noise ratio in T2-enhanced steady state gradient echo sequences. This type of sequence is rarely used in clinical practice.
DESS (Double Echo Steady State)
The DESS sequence is a combination of FISP and PSIF sequences . The FID signal improves spatial resolution and the Hahn echo improves T2-weighting.Interest: high-resolution T2 -weighted 3D gradient echo imaging.
Balanced gradient echo
Type of sequence
|
Philips
|
Siemens
|
GE
|
Hitachi
|
Toshiba
|
Balanced GE
|
Balanced
FFE |
True FISP
|
FIESTA
|
BASG
|
True SSFP
|
The steady state is perturbed by the phase shift linked to rapid flows. The shift alters phase encoding and degrades the quality of the image.
By applying balanced and symmetrical gradients in the 3 spatial directions, phase shifts induced by flow at constant speed are nulled. Balance indicates equal quantities of positive and negative lobes The recorded signal simultaneously cumulates the free induction signal (FID) and that of the spin/ stimulated echo.
The contrast varies according to T2/T1 ratio.
With these sequences, we obtain ultrafast (roughly one second per slice), and robust imagery with a liquid / tissue contrast and an excellent signal to noise ratio.
High TR is likely to produce band artifacts.
CISS
Combination of 2 True FISP acquisitions, with and without excitation pulse phase alternation, to eliminate band artifacts .Interest: high resolution T2 -weighted 3D gradient echo.
Echo planar (EPI)
Type of sequence
|
Philips
|
Siemens
|
GE
|
Hitachi
|
Toshiba
|
SE - Echo planar
|
SE-EPI
|
EPI SE
|
SE EPI
|
SE EPI
|
SE EPI
|
GE - Echo planar
|
FFE-EPI
TFE-EPI |
EPI Perf
EPIFI |
GRE EPI
|
SG-EPI
|
FE-EPI
|
- an excitation pulse, possibly preceded by magnetization preparation
- continuous signal acquisition in the form of a gradient echo train, to acquire total or partial k-space (single shot or segmented acquisition)
- readout and phase-encoding gradients adapted to spatial image encoding, with several possible trajectories to fill k-space (constant or intermittent phase encoding gradient, spiral acquisition
Preparation and contrast
Contrast in echo planar sequences is determined by excitation pulse and possible magnetization preparation. The various possibilities include:- GE-EPI: single RF excitation pulse, with no preparation -> T2* weighting
- SE-EPI: pair of 90° - 180° pulses (spin echo type) -> T2 weighting
- IR-EPI: 180° inversion pulse to prepare magnetization then RF excitation pulse -> T1 weighting
- DW-EPI: preparatory pattern for diffusion weighting
GE-EPI
SE-EPI
DW-EPI
Gradients and k-space filling
In the case of spiral k-space filling, phase encoding and readout gradients have a sinusoidal growing envelope.
In all cases, the continuous readout signal imposes k-space regularisation before the image can be reconstructed. The Fourier plane matrix values are calculated by mathematical interpolations that are more or less complex depending on the filling trajectory used.
Echoplanar sequences demand intense, high-performance gradients (for fast signal readout), with short ascent times (because of frequent gradient-switching).
Artifacts
The artifacts in echoplanar sequences are linked to:- sensitivity to magnetic susceptibility, which can be reduced by using segmented rather than single-shot sequences, at the cost of increasing scan time
- gradient imperfections (particularly induced currents) which perturb spatial encoding, leading to ghost images
- the narrow readout bandwidth in the phase-encoding direction, provoking chemical shift artifacts in this direction, thus requiring suppression of the echoplanar fat signal
No comments:
Post a Comment