Table of contents
1. Pulse sequence
pulse isA signal that occurs for a short period of time throughout the signal period relative to a continuous signal, there is no signal for most of the signal period.
The sequence has a certain bandwidth and a certain amplitudeOrganic combination of radio frequency pulse and gradient pulse.
The signal strength of an MR image depends on the way the radio frequency pulse is transmitted , the way the gradient magnetic field is introduced , and the way the MR signal is read . A series of radio frequency pulses, gradient pulses and signal acquisitions designed for different imaging purposes are arranged in a certain time sequence called a pulse sequence.
For the convenience of subsequent understanding, the definitions of pulse and pulse sequence are given here: the step of generating a magnetic resonance image data is usually called a pulse sequence .
Original Fourier imaging
↓ (modify t1 as a fixed amount, Gy as a step value)
spin-warp Fourier imaging
↓ (acquire echo signals instead of free induction attenuation signal FID) (avoid front-end error and delay sampling)
back Wave technology (usually with some signal loss)
Spin echo (SE) sequence、Reverse Recovery (IR) Sequence、
Gradient echo (GE) sequence、stimulated echo sequence
2. Spin Echo (SE) Pulse Sequence
2.1 Spin echo pulse sequence
Layer selection gradient Gs, readout gradient (frequency encoding gradient) Gr, phase encoding gradient Gp (programmable step value), echo time TE, repetition time TR
Assuming that the magnetic field is inhomogeneous, after the 90° pulse, after a certain time Ti, the transverse magnetization vector will be separated ; at this time, a 180° refocusing pulse is applied to reverse the phase of the transverse magnetization vector, and then aftersame Ti time,Can achievephase reunion
Echo time TE (Time of Echo) refers to the time interval between the radio frequency pulse and the corresponding echo . TE determines the acquisition time of the echo, and also determines the attenuation degree of the tissue macroscopic transverse magnetization vector Mxy.
The signal peak is already present after the 90° RF pulse, but what is needed is an echo, i.e. a convoluted signal peak
. Therefore,Use 180° rephasing pulses to get echoes.
Gs first blockthe top halfEasy to understand, it is used to make the transverse magnetization of different layers have different precession frequencies, and then realize layer selection ;the second halfThe FID signal dumped after the 90° RF pulse decays to zero as soon as possible to shorten the echo time TE, thereby improving the imaging speed .
The second block of Gs is used to compensate the out-of-phase after the 90° pulse, so that the echo peak value is maximized .
The first block and the second read gradient leaf cooperate to complete the frequency encoding.
2.2 Signal-to-noise ratio, difference-to-noise ratio
Signal-to-noise ratio , as the name implies, is the signal-to-noise ratio
. Depending on the echo sequence used, the definitions of signal-to-noise ratio and difference-to-noise ratio are also different.
Signal-to-noise ratio SNR : It is related to the number of excitations, the number of sampling points, the number of encoding steps, etc., and is used to measure the quality of the obtained signal .
In order to obtain better quality images , but not limited toReduce the thickness of the sampling slice、 Reduce pixel size (reduce field of view or increase matrix)The spatial resolution can be improved by means of other methods, but they are limited by the signal-to-noise ratio; the number of excitations can also be increased, but this means a longer scan time or a larger field strength B0 .
Difference-to-noise ratio CNR : signal-to-interference-to-noise ratio, which is the ratio of the intensity of the received useful signal to the intensity of the received interference signal, and is used to evaluate the ability of MRI to detect low-contrast lesions.
2.3 Sampling
Frequency encoding Kx, phase encoding Ky, number of columns (number of sampling points) Nx, number of rows (number of steps of phase encoding) Ny .
Fourier row (view) : take one echo to fill one row of the data matrix, and the time interval between Fourier rows is the repetition time TR .
The collection of data is a process of filling the matrix , and each point collected (Here is called matrix element) when the sampling frequency needs to satisfy the Nyquist theorem , considering that the gradient field and the B1 field of the RF body coil cover the entire object or section,The actual highest Larmor frequency in the object is greater than the highest signal frequency, it is necessary to ensure that the sampling frequency is greater than twice the highest frequency of the signal . If it is equal to or even lower than twice the Nyquist frequency, the sampling data will have false low-frequency parts. This phenomenon isAliasing。
Oversampling is generally used to solve the aliasing problem . As the name suggests, namelySet the sampling frequency twice as high as the highest Larmor frequency of the object, and discard the extra part of the image after image reconstruction。
The K-space data matrix obtained by sampling and filling the original data signal. The acquisition here also needs to be explained.
Due to the influence of T2 relaxation , the center was collected first, and then both sides were collected to weaken the influence of T2 . The order in which the Fourier rows are acquired, sequential acquisition levels may occurlevel interference, the thickness of the sampling layer also has certain requirements , which are temporarily shelved. Data reconstruction is performed on the K space, and its values are displayed in grayscale values, which is the lower right figure, and the reconstructed image can be obtained after Fourier transformation and other operations.
2.4 Improved spin echo variant sequence
Increase the number of excitations to further improve the signal-to-noise ratio, and use multiple phase encoding steps with the same 180° pulse to obtain multiple spin echoes . The number of echoes increases , equivalent to the acquisitionData increase in phase, can naturally improve the signal-to-noise ratio . But it does not increase infinitely, and the echo will attenuate due to T2 relaxation, which also causes the subsequent echo to gradually become smaller.
The concept of weight is relatively clear, and weighting is to highlight the meaning of part of the data. Taking the double-echo sequence as an example, further analyze it. The TE of the first echo is short, and the spin density imaging can be obtained ; while the echo time of the second echo is longer, which can be approximated as T2, and the T2-weighted image can be obtained .
Likewise, a single RF excitation applies multiple 180° pulses to generate multiple echoes .
The gradient encoding step Gp applied each time is different, so multiple echoes correspond to multiple different gradient encoding steps; and after each echo is collected, a wrapping gradient in the same direction as Gp is applied ,Eliminates the influence of phase gradually no longer converging caused by the difference in phase encoding gradient. The signal-to-noise ratio is improved , and the scan time is also greatly shortened .
3. Reverse Recovery (IR) Pulse Sequence
3.1 Reverse recovery pulse sequence
The reverse recovery pulse sequence consists of180°x-90°-180°y pulseandThree orthogonal gradient pulses (layer selection, phase editing, frequency editing)Composition, 180°x pulse means add 180° pulse on x-axis, 180°y pulse means add on y-axis.
The echo obtained by 180°x is opposite to the original FID signal, and the echo obtained by 180°y is in the same direction as the original FID signal.
As previously discussed, always apply a 90° pulse to let Mz fall, then apply n 180° pulses, and then wait for Mz to recover , which causeswaste of time. You can apply a 90° pulse after applying a 180° pulse to accelerate the waiting time ; you can also fine-tune the RF during the waiting time , sequentially excite the K1 plane and then collect the data of the ath row, and then excite the K2 plane to collect the data of the ath row... Until the same row of data of each layer is collected, the first excited K1 plane Mz has been restored.
Combining the fast spin echo fSE with the multi-slice scanning MSE , the obtained multi-slice fast SE sequence can further shorten the acquisition time, which will not be repeated here. The issue of RF power deposition needs to be considered, but there is no need to design the pulse sequence, so it will not be expanded.
(Most of the RF power is consumed by the patient, and a small part is consumed by the RF coil, which causes the coil to heat up.. The power consumed by an RF pulse is proportional to the effective volume of the coil, and proportional to the square of the Larmor frequency)
first apply a 180° pulse to reverse the Mz of the selected layer (Mz=M0), after the 180° pulse is applied, 90 ° The period of time before pulse applicationMz decays with T1 time constant and recovers towards the original M0. Therefore, the choice of T1 determines the direction in which Mz nutates to the transverse plane when a 90° pulse is applied . Afterwards, a 180° pulse flip on the y-axis is performed, and the principle is the same as that of the previous spin echo sequence SE to obtain an echo.
3.2 Improvements on the reverse recovery pulse sequence
Image contrast : In MRI, there are three sources of intrinsic tissue contrast: proton density N(H), T1, and T2 .
By selecting the appropriate pulse sequence , timing parameters , slice thickness , matrix and appropriate field of view (FOV) , theThe contrast or gray value (signal difference) of the image reflects the intrinsic contrast of the tissue, to distinguish lesions from normal tissues .
For the combination of normal IR and fast spin echo fSE, multiple 180° pulse sequences are added to the reverse recovery pulse sequence, and multiple pairs of equal large reverse gradient pulse sequences are applied to accelerate the recovery of Mz and simultaneously acquire multiple echo.
But affected by RF power deposition . This is multiple acquisitions at one level, and the multi-level acquisition just now can also be achieved, which is a multi-level fast IR sequence.
4. Gradient echo (GE) pulse sequence
4.1 Basic concept of GE sequence
First use the first gradient pulse to make the nuclear magnetization out of phase , and then use the second same width, same amplitude (or equal area) and reverse gradient pulse to make the magnetization phase together , so thatgenerate echo。
Small-angle dumping, the horizontal component My=M0sinθ, leaving the vertical component Mz=M0cosθ
same withlose phase and reuniteHowever, since the 180° pulse is not applicable to the gradient echo GE sequence, but the echo is formed by gradient inversion , the 90° RF pulse is no longer used this time, and the inclination angle θ of the α pulse can be selected to be very small. It can be seen from this formula that if θ is small enough, the longitudinal component left is quite considerable, reducing the recovery time TR of Mz ,Further reduce overall scan time. The reduction of TR here may become the source of artifacts, we will discuss it next time.
At the same time, notice that the horizontal component decreases, the signal decreases, and the signal-to-noise ratio SNR core difference-to-noise ratio CNR also decreases. You can define the signal-to-noise ratio per unit time, the difference-to-noise ratio per unit time, the horizontal scanning time, the signal-to-noise ratio, and the difference-to-noise ratio. relationship between those.
4.2 3D Imaging
Three-dimensional imaging is made possible by RF imaging due to its high speed and small RF power deposition (small angle dumping) .
Gs is used for block selection and also used for the first phase encoding, Gp is used for the second phase encoding, and Gr is the read gradient.
A very short sequence repetition time TR is allowed , the total imaging time is greatly shortened, and the signal-to-noise ratio is greatly improved; but its tolerance to artifacts is poor , and when the slice is thicker, it is affected by truncation artifacts.
The steps of three-dimensional imaging are specific There are two steps, the first step is to cut the tissue , and use Gs to pull the selected magnetization to the xy plane. The second step is to cut the excited block , which is calledlayer coding gradient; Then use the phase encoding gradient Gp and the frequency encoding gradient Gr to distinguish the two dimensions of the plane.
5. Coherent steady-state GE pulse sequence (GRASS)
5.1 Reunited phase of residual transverse magnetization
Because the phase encoding gradient will cause the phase divergence of the magnetization along this direction ,Adding a constant amplitude reverse gradient can fully compensate for this dispersion, so that the tissue and components with long T2 are displayed as high signal, increasing the image contrast.
Has the effect of imaging blood vessels, spinal cord, joints to determine if blood vessels are open or if there is fluid in an area.
Sensitive to flow, good vessel images can be obtained.
The short TR value of the GE sequence makes the transverse magnetization residual before the next measurement starts after the end of the echo measurement .
To deal with this transverse magnetization that has not decayed to zero,Apply a gradient pulse of equal step and reverse direction after the end of the measurement, turning this transverse magnetization to the z-axis compensates for the phase divergence caused by the first phase encoding gradient.
5.2 SSFD Double Echo and True FISP Sequence
Steady-state free precession (SSFD) : When the radio frequency is continuously excited at very short repetition time (TR) intervals (TR<<T2), a stable mixed echo phenomenon is produced . The resulting echoes are located on either side of the excitation pulse, with the free induction decay on the right and the excitation echo signal on the left.
Double echo SSFP=FISP+PSIF .
In SSFP-FID and SSFP-Refocused, the excitation echo is collected once, and the two echoes adopt the same phase encoding to fill the same phase encoding line in K space .
The steady-state free precession SSFP double-echo sequence isOrdinary steady-state free precession sequence FISPandFISPcombination.
The center of the gradient echo generated by the original FID signal is at t1, the center of the gradient echo generated by SSFP-echo is at t2, and the distance between the two waves is Δt=t2-t1=TR/2. Two kinds of echoes are collected at the same time, the SNR is higher, and the T2 weight is heavier.
6. Irrelevant GE sequences
6.1 Destructive Gradient Echo (sGE) Sequence
The methods to generate gradient echo and exclude transverse magnetization are as follows: one is to useTR is long enough, so that the residual transverse magnetization vector is completely lost at the beginning of the next sequence; the second isIn the case of short TR, the residual transverse component and its phase relationship are corrupted after the signal measurement.
After each data collection, a gradient pulse is applied in the layer selection direction to destroy the residual transverse component .
Note that in order to avoid the establishment of coherence between the residual transverse components at different acquisition times, the magnitude of the destruction gradient used needs to be varied for each excitation . This approach allows very short TRs without saturation.
For destructive gradient echo (sGE) imaging, there are the following rules:
Relatively long TR, small excitation angle θ and short TE can obtain spin density-weighted images; relatively long TR, small excitation angle θ and long TE can obtain free induction decay-weighted images. T1-weighted images can be obtained with short TR, large excitation angle θ, and short TE。
7. Ultrafast FLASH pulse sequence
1. Ultrafast FLASH imaging with spin density weighting
2. T1-weighted reverse recovery (IR) ultrafast FLASH imaging
3. T2-weighted ultrafast FLASH imaging
4. Chemical shift selective saturation ultrafast FLASH imaging
5. Ultrafast FLASH imaging of NMR spectrum
8. Stimulated echo pulse sequence
The function of the second RF pulse is to store the magnetization in the longitudinal direction . Between the second and third RF pulses, each color spin group remembers its own precession phase, so this period is the magnetization "storage time" (TM).
The third RF pulse is called " readout pulse ", which pulls the magnetization stored in the longitudinal direction back to the transverse plane and begins to experience T2 relaxation. Since the precession phase accumulated during τ1 is remembered, it passes through τ1 Time forms the stimulated echo.