Pulse Sequence Part I

• Without pulse sequences we can‟t do MRI. Our life depends on it in terms of which kind of image contrast we want to see or, even, which kind of pathology we want to detect.
• Understanding what a pulse sequence is and how it influences the image is vitally important.
• A pulse sequence is a sequence of events, which we need to acquire MRI images.
• These events are: RF pulses, gradient switches and signal collecting.
• Each sequence will have a number of parameters, and multiple sequences are grouped together into an MRI protocol. 
• Multiple sequences are usually needed to adequately evaluate a tissue, and the combination of sequences is referred to as a MRI protocol.
• This figure   shows a “sequence diagram” in which the order of the events are shown schematically.
  Lets start of with (1) switching on the Slice Select gradient (G_SS). Simultaneously (2) a 90º RF-pulse was given to „flip‟ the net-magnetization into the X-Y plane. Then (3) the Phase Encoding gradient G_PE was switched on to do the first phase encoding. Then (4) the Frequency Encoding or Read Out gradient (G_RO) was switched on during which (5) the signal, the Free Induction Decay (FID), was sampled. 
  
  
  • This is a very simple and basic sequence.
  • We also saw that the signal dies out very quickly.
  • In the early days that was a problem. The hardware could not be switched quick enough to sample the entire signal. They could only sample the last part of the signal when most of the signal was gone.
  • The resulting image showed it, It was signal starved. In order to improve the amount of signal the engineers came up with a brilliant solution.
  
  A pulse sequence is generally defined by multiple parameters, including:
  Different combinations of these parameters affect tissue contrast and spatial resolution.
  These parametera are:
  
  • Time to echo (TE)
  • Time to repetition (TR)
  • Flip angle
  • Field of view and matrix size
  • Inversion pulse(s)
  • Spoiler gradient(s) (crusher gradients)
  • Echo train length (ETL)
  • The spatial acquisition of k-space
  • 3D acquisition vs. 2D acquisition vs. multiple overlapping slab acquisition
  • Post contrast imaging with gadolinium contrast agents
  • Diffusion weighting (b values)
  
  
  MRI sequences can be grouped in a number of ways. Probably most accurately they are grouped according to the type of sequence (e.g. Spin echo, or inversion recovery etc..)
  However for non radiographer or non radiologists another way of grouping sequences is by general image weighting (e.g. T1 or T2) and additional features (e.g. Fat suppressed or gadolinium enhanced).
  
  
  
  Pulse sequences can be broadly grouped as follows:
  • Spin echo sequences
  • Inversion recovery sequences
  • Gradient echo sequences
  • Diffusion weighted sequences
  • Saturation recovery sequences
  • Echo-planar pulse sequences
  • Spiral pulse sequences
  
  

  Spin Echo (SE) Sequence

  • Spin-echo pulse sequences are one of the earliest developed, most commonly   and still widely used (in the form of fast spin-echo) of all MRI pulse sequences.
  • The pulse sequence timing can be adjusted to give T1-weighted, proton density, and T2-weighted images.
  • Dual echo and multiecho sequences can be used to obtain both proton density and T2-weighted images simultaneously.  
  • During the process of spin echo sequence, a  90^o pulse is first applied to the spin system.  The  90^o degree pulse rotates the magnetization down into the X'Y' plane. The transverse magnetization begins to diphase.
  • It is because of this dephasing that the signal drops like a stone.
  • Ideally, we would like to keep the phase coherence because this gives the best signal. The brilliant solution the engineers came up with is this: a short time after the 90º RF-pulse a second RF-pulse is given.
  • At some point in time after the 90^o pulse, a 180^o pulse is applied. This pulse rotates the magnetization by 180^o  about the X' axis  and avoid signal loss caused by local inhomogeneity in the magnetic field
  • The 180^o pulse causes the magnetization to at least partially rephase and to produce a signal called an echo.
  • When all the spins are rephased the signal is high again, and when we make sure we sample the signal at this instant we‟ll have a much better image.
  • Notice that the 180º rephasing pulse is exactly in the middle of the 90º pulse and the echo.
  • Each echo is used to fill a single horizontal line in a raw data matrix called k-space.
  • The process is then repeated by applying a second 90° pulse and so on.
  • Once all lines of the raw data matrix have been filled, sufficient information is available to calculate an image.
  • The two variables of interest in spin echo sequences are the repetition time (TR) and the echo time (TE).
  • In order to allow for recovery of the longitudinal magnetization after a 90° pulse, TR values should be sufficiently long.
  • Spin echo MRI is a rather slow technique that cannot be used for imaging during breathholding.
  • All spin echo sequences include a slice selective 90-degree pulse followed by one or more 180 degree refocusing pulses as shown in the diagrams.
  This  figure  show it better.
  
  
  
   A timing  diagram showing the relative positions of the two radio frequency pulses and signal
  
  
  
  
  
  
  One horizontal line of the raw data matrix (k-space) is filled after each 90° excitation pulse. The scan time is determined by the length of the TR.
  As with everything in MRI, the spin-echo sequence is a compromise:

  Advantages 

  • The signal is strong
  • Compensation for local field inhomogeneities: less artifacts.

  Disadvantages 

  • It takes time to do the rephasing step. This will increase the total scan time.
  • It increases the amount of RF one has to put into the body (not that it‟s dangerous, but there are certain limits).
  In spite of the increased scan time and the amount of RF the spin-echo sequence is widely used and has become the routine sequence in MRI.
  
  

  TR (Repetition Time)  

  • As stated before, the whole process must be repeated as many times as the matrix in the phase encoding direction is deep.
  • TR is the time between two 90º excitation pulses.
  • In regular SE sequences the TR can be anything in the range of 100 to 3000 milliseconds.

  TE (Echo Time)  

  • This is the time between the 90º excitation pulse and the echo.
  • The TE can be anything in the range of 5 to 250 milliseconds.

  FA (Flip Angle) 

  • Refers to the amount of degrees the net-magnetization is flipped into the X-Y plane.
  • It has nothing to do with the 180º rephasing pulse.
  • The FA in a normal SE sequence is always 90º, however, in modern SE sequences this can be varied as well.
  • FA‟s of 70º and 120º are quite common, although the FA can be choosen between 1º and 180º.

  Multi-slice sequence

  • An imaging technique in which the repetition time  (TR) is utilized for acquiring additional slices in other layers or planes, in differentiation for 2D techniques where every repetition time is used for single slice.
  • The maximum number of slices of a pulse sequence depends on the repetition time.
  • The TR is necessary to allow sufficient time for the T1 relaxation to complete in order to have enough Mz to give a signal when it is flipped by the 90° pulse.

  An example: Let us assume we are making a scan of the brain.

  • A typical brain scan contains 18 slices
  • TE of 30 milliseconds
  • The TR is 540ms
  • A matrix size of 256 x 512 (256 phase encoding steps are required per slice)

  The scan time is:

  • TR x PE steps x Number of slices / 60,000
  • 540 x 256 x 18 / 60,000 = 41.4 minutes

  Considering the TE is only 30ms, this is a very long scan with a lot of dead time in which no signal is being created. We can use this dead time by selecting another slice and starting a cycle, then selecting a third slice and starting a cycle etc.

  After 540 ms it is time to start the second cycle for the first slice. In 540 ms we can scan 18 lines of 18 different k-spaces.

  Now we just need to repeat this enough times to get every line of every k-space (i.e. multiple by the number of phase encoding steps).

  Recalculating the scan time gives us:

  • 540 x 256 / 60,000 = 2.3 minutes

  Now, that‟s a scan time we can live with. Especially when after this time the whole brain is imaged.

  This technique is used in nearly every scan to make the scan times shorter.

  This trick is called Multi-Slicing. It is used in just about every scan we make so as not to waste any time. Just imagine when multi slicing would not be possible. MRI would be even slower than the Internet
  
  
  

  Multi Echo Sequence

  • So far we used one echo in our sequence we can acquire more echoes in one cycle. 
  • By repeating the sequence we filled one k-space and this generated one image. It is, however, possible to acquire more echoes in one sequence.
  
  
  
  
  
  • When an 180º rephasing pulse is applied we saw that the spins are rephased until they were all in phase again, but it doesn‟t stop there. What happens next is that the spins will, once again, start to dephase due to T2 properties.
  • So, once more, we can apply a (second) 180º rephasing pulse to rephase the spins until they create a second echo. When we sample the second echo we put it in a second k-space. When all the lines of both k-spaces have been filled we end up with two different images.
  • The second image has a different contrast because the TE is different.
  • The first image is a so-called Proton Density  image, while the second image is called a T2 image.
  • PD weighted uses a short TE of 15ms
  • T2 weighted uses a long TE of 1000-3000ms
  • If we look at the images in the figure above, we see this contrast difference. The cerebro-spinal-fluid (CSF) in the PD image is dark, while it is bright in the T2 image.
  • This type of sequence is called a Double-Echo Spin Echo sequence or, more commonly used, a SE T2 sequence.
  • This technique can be combined with Multi-Slicing bearing in mind that for a T2 sequence a very long TR is used (2000 ms or more).
  • The long TR is necessary to allow for complete T1 relaxation of water.
  
  

  Image Contrast

  • Before we move on to other pulse sequence techniques it is a good time to discuss image contrast.
  • We have seen that there are two relaxation processes, T1 and T2, going on at the same time. The image contrast is highly dependent on these relaxation processes.
  • Image contrast depends on how much of each process we allow to happen.
  • An example might be useful here:
  
  

  T1 Contrast

  • T1 weighting tends to have short TE and TR times
  • Assume we scan with the following parameters: TR 600 and TE 10.
  • We allow for T1 relaxation to take place for 600 milliseconds and, more important, T2 relaxation only for 5 milliseconds (10÷2).
  
  
  • When we look at figure A, we see that after 5 ms. hardly any dephasing has taken place. We receive a lot of signal from all tissues. The image contrast is, therefore, very little influenced by T2 relaxation.
  • In figure B we see that after 600 ms. not all tissues have undergone complete T1 relaxation. Fat is nearly there, but CSF has still a long time to go.
  • So, for the next excitation the net magnetization vector of the CSF spins, which can be flipped into the X-Y plane is small. This means that the contribution from CSF to the overall signal will be small too.
  • In short, the image contrast becomes dependent on the T1 relaxation process.
  • In the final image CSF will be dark, fat will be bright and gray matter will have an intensity somewhere in between.
  In this case we say that the image is “T1 weighted” because the contrast is more dependent on the T1 relaxation process.
  

   

  Summary

  In T1 Weighted image:
  
  • Fat bright
  • Fluid dark
  • TR: short
  • TE: short
  
  
  

  T2 Contrast

  • In another example we use the following parameters: TR 3000 and TE 120.
  • Now we allow T2 relaxation to happen for 60 ms. (120÷2). As we can see from the figure A below most of the tissues have dephased and won‟t produce that much signal.
  • Only CSF (water) has still some phase coherence left. Here the TE is the dominant factor for the image contrast.
  
  
  • Figure B shows that practically all tissues have undergone complete T1 relaxation.
  • The long TR of 3000 ms does not contribute much to the image contrast. The 3000 ms. are only needed to allow CSF to recover completely before the next excitation.
  • In our image we‟ll see CSF bright, while the other tissues show up in various shades of gray.
  • In this case we say the image is “T2 weighted” because we allowed for T2 to happen for a “long” time.
  
  

  Summary

  • TR: long
  • TE: long
  • flip angle: less important than with T1 weighting
  • Fat: intermediate-bright
  • Fluid: bright
  

   

  Proton Density Contrast

  • There is one more type of image contrast called Proton Density.
  • Now we choose the parameters: TR 2000 and TE 10.
  • Again we allow T2 relaxation to happen for only 5 ms., which means that T2 relaxation contributes very little to the image contrast.
  • With a TR of 2000 ms. the net magnetization of most tissues will have recovered along the Z-axis.
  • The image contrast in PD images is neither dependent on T2 relaxation, nor T1 relaxation.
  • The signal we receive is completely dependent on the amount of protons in the tissue: few protons means low signal and dark in the image, while many protons produce a lot of signal and will be bright in the image.
  • It is important to understand that all images have a mix of T1 and T2 contrast. It just depends on how much T2 relaxation one does allow to happen. In SE sequences the TR and TE are the most important factors for image contrast.
  
  
  • This figure  shows examples of various image contrasts; T1 weighted, Proton Density and T2 weighted.
  • Notice the differences in signal intensity of the tissues. CSF is dark in T1, gray in PD and bright in T2.
  • It is quite amazing that only two parameters, TR and TE, can create so many different contrasts, even more so when we choose different values for these two parameters but there are two more parameters, which influence image contrast.
  
  
  This figure shows in a diagram how TR and TE in a SE sequence are related in terms of image contrast.
  
  • A short TR and short TE gives T1 weighted contrast.
  • A long TR and a short TE gives PD contrast.
  • A long TR and long TE gives T2 weighted contrast.
  
  

  Summary of SE-weighted Operator Parameters

  

  
  

  When To Use Which Contrast

  • With all these types of contrasts to choose from, one can ask which contrast to use in a particular situation.
  • This is a valid question, but the answer is not so easy.
  • Certain pathologies show better on PD weighted images than on T2 weighted images, while others show better on T1.
  • In general, though, we can stick to the following rule:
  • For a clear delineation of anatomical structures a T1 weighted or, even better, an IR sequence is the best choice
  
  
  Anatomy
  
  
  • For pathology, PD weighted or, preferably, T2 weighted contrast is used. The reason being that most pathology produces water (edema), which shows bright on T2 weighted images.
  
  
  Pathology
  
  
  • Another option would be to inject a MR sensitive contrast medium, such as Gadolinium-diethylenetriaminepentaacetic acid (Gd-DTPA or Gad for short ),
  • In this case you would scan a T1 weighted sequence because Gd-DTPA shortens the T1 relaxation time of tissue and shows, therefore, bright on a T1 weighted image. (Gd-DTPA does not show on T2 weighted images).
  • In practice almost always both T1 weighted (with and without Gd-DTPA) and T2 weighted sequences are scanned in one or more planes, to ensure optimum visualization of the pathology at hand.
  
  

  Turbo Spin Echo (TSE or FSE) Sequence

  • Although techniques such Multi-Slicing can be used to reduce the scan time, a regular SE T2 sequence can still take up to 12 minutes to acquire.
  • During this time a practical problem arises: patient movement. It is very difficult to lie still for such a long time. And yet, that is necessary otherwise the image would be totally useless due to motion artifacts.
  • In order to reduce the scan time a very clever German chap called Henning came up with the Turbo-Spin-Echo (TSE) sequence (also known as Fast-Spin-Echo (FSE)).
  • A fast spin echo imaging sequence is a multi-echo spin-echo sequence where different parts of k-space are recorded by different spin-echoes.  
  • The benefit of the technique is that a complete image can now, as was shown in thie figure below example, be recorded in one of the time.
  
  
  
  
  
  • The TSE sequence also makes use of the multi-echo principle as shown in the figure above. After the 90º pulse a series of seven 180º pulses are given. Each 180º pulse generates an echo.
  • k-Space is divided into 7 segments and each echo fills one line in each segment.
  • One, usually T2, image is reconstructed.
  • The advantage of this technique is clear: a scan time reduction factor of 7.
  • Compare these scan times:
  Normal (Regular) spin echo = TR x no. GPE (MX_PE) x number of slices
  Turbo spin echo = TR x no. GPE (MX_PE)  x number of slices / ETL
  
  • Regular SE: TR 3000, TE 120,  MX_PE 256 works out to 3000 x 256 = 12.8 minutes.
  • TSE: TR 3000, TE 120,  MX_PE 256 and 7 echoes: (3000 x 256) ÷ 7 = 1.8 minutes.
  • You can understand that this type of sequence is very useful. Funnily enough many radiologists were not so keen to use this sequence. They were used to seeing a specific T2 contrast when scanning the brain.
  • The FSE image has a mix of contrasts. The figure above helps us to understand this. As we know signal and contrast information is stored in the centre of k-space. In our example you can see that the 4th echo, as well as parts of the 3rd and the 5th echo, is put in the centre of k-space.
  • Because each echo is at a different time, each echo will carry different contrast information. The result is an image with a mix of contrasts.
  • Another negative aspect is that the image will show artifacts, which are specific for this type of sequence.  
  • All in all in it took some time before TSE was widely accepted. There are radiologists, however, who still don‟t use TSE because of these drawbacks.
  • A number or series of echoes as used or created in TSE is called an Echo Train Length (ETL). One can choose how many echoes one would like to use or create. In our example we used an Echo Train Length (ETL) of 7 but an ETL of 212 is also possible.
  • It is also possible to create two images out of an Echo Train Length. All we need is two k-spaces. For example, with an ETL of 14, one can use the first 7 echoes for a PD image (1st k-space) and the last 7 echoes for a T2 image (2nd k-space). This is called a Double-Echo TSE sequence or PD/T2 TSE sequence.
  • The TE measured is taken to be the echo created when the GPE is zero and is called the "effective TE (T_eff)".
  N.B  The phase encoding gradient is reversed prior to the next 180° RF pulse to rephase the spins.
  
  

  Advantages

  • Very fast - useful for MR angiography in which very fast scan times are needed.
  • Can create two images of different contrasts by filling two different k-spaces. E.g. if we have an ETL of 14, we can use the first 7 echoes for a PD image (first k-space) and the last 7 echoes for a T2 image (second k-space). This is called a Double-Echo TSE sequence
  
  

  Disadvantages

  • Only really able to achieve heavily T2 weighted images
  • Mix of contrasts: Each echo that fills a different line of k-space is at a different time and, therefore, a different contrast.
  
  

  

  

   

  Fast Advanced Spine Echo or HASTE Sequence

  • We can take TSE a step further and fill an entire k-space in one cycle. .
  • In the Fast Advanced Spin Echo (FASE) sequence an ETL of 212 is used. This already results in ultra short scan times (reduces the scan times significantly).
  • Furthermore, we only really need to fill up just over half of k-space (i.e. 212 rows). We can then use a Half Fourier Imaging (HFI)  to extrapolate the rest of k-space and complete the image.
  • The combination of 212 echoes and HFI results in scan times, which are only a fraction of a regular SE sequence. The figure below show  how it works.
  • The very late echoes are put in the centre of k-space (heavily T2 weighted) which results in an image that only shows free water.
  • This kind of sequence is used for MR-Cholangio-Pancreatography (MRCP study).
  
  
  • Each echo of this 212 ETL sequence fills one line in k-space. This fills k-space for slightly more than 50%. The rest of k-space is filled with zeroes (no data). The stunning bit is that you need only one repetition to create an image.
  • Notice that only the last (very late) echoes are put in the centre of k-space resulting in an image, which only shows free water (bile and water in the intestines as the image shown in figure above). 
  
  
  
  
  

  Gradient Echo (GE) Sequence

  • A second group of sequences are the Gradient Echo sequences. Also with this type of sequence an echo is rebuilt from the FID.
  • Gradient echo sequences (GRE) are an alternative technique to spin echo sequences, differing in the way the echo is formed from it in two principal points:
  1. Utilization of gradient fields to generate transverse magnetisation
  2. Flip angles of less than 90° (Where a Spin Echo sequence uses an 180º rephasing pulse to rephase the spins, the Gradient Echo sequence uses a gradient polarity reversal). This will lead to a potential shortening of the echo time and repetition time.
  • Gradient echo sequences, is also known as gradient-recalled echo or fast field echo (FFE), employ the gradient coils for producing an echo rather than pairs of RF pulses.
  • This is done by first applying a frequency-encoding gradient with negative polarity to destroy the phase coherence of the precessing spins (dephasing). Subsequently, the gradient is reversed and the spins rephase to form a gradient echo
  • Because gradients do not refocus field inhomogeneities, GRE sequences with long TEs are T2* weighted (because of magnetic susceptibility) rather than T2 weighted like SE sequences.
  • This feature of GRE sequences is exploited- in detection of hemorrhage, as the iron in Hb becomes magnetized locally (produces its own local magnetic field) and thus dephases the spinning nuclei. 
  
  
  In  Gradient echo sequences (GRE) the following are applied:
  
  • RF pulse applied
  • Slice-select gradient applied
  • Phase-encoding gradient applied
  • Frequency-encoding gradient applied
  1. A negative GFE is applied. The spins dephase, some faster than others.
  2. The positive GFE is applied. The spins start to rephase until they are again in phase and a signal is created - the Gradient Echo

  

  1. Slice selecting with G_SS.
  2. Send excitation pulse.
  
  
  3. Phase encoding.