I never wanted to publish about RF sequences before, simply because new RF sequences will be implemented on the different MRI systems faster than you can hit the print-button. But finally, now I’m having these modules on-line, I can update them constantly! I can not cover all RF sequences but I try to cover the most-important and most-used ones!
Regarding all the acronyms and abbreviations I would like to suggest: explore the internet: Glossary of MRTerms-ACR: www.acr.org › Files › MRGlossary
But they only cover Siemens, GE and Philips. The Siemens MRI Acronym list from 2019 has the acronym list of Siemens, GE, Philips, Canon and Hitachi: www.edumedag.com › wp-content › uploads › 2019/05
https://wissen.science-and-fun.de/chemistry/spectroscopy/abbreviations-and-acronyms-use-in-nmrmri/ : here you can also add terminology.
Many great online resources for lifelong MRI education are available like on the websites of the ISMRM, ESMRMB and ESOR. Find your way: explore!
Please be sure to check the website of your MRI manufacturer to see the acronym comparisons.
For medical abbreviations: https://radiopaedia.org/articles/medical-abbreviations-and-acronyms-a
To the best of my knowledge there is no website covering all the RF sequences implemented on the different MRI systems. Please let me know if there are different RF sequences implemented on your MRI system; I am open to suggestions!
There are always new technologies in development and scientific “energies” underway: try to stay updated!
Although I am more going into the RF sequences of the High field MR systems, I am very much aware of the fact that the Lower field MR systems are gaining in numbers. And… the basics, the implementation of the RF sequences are still rather similar!
(Ultra) High Field – (Ultra) low Field MRI systems
A High field superconducting MRI magnet has a field strength of 1 Tesla or higher. Currently they are routinely used as 1.5 and 3T MR systems since the 80ties.
In July 2017 an Ultra high field magnet (11.7 Tesla), was installed by the NeuroSpin institute, a research center dedicated to human brain imaging at the Commission for Atomic Energy and Alternative Energies (CEA) Paris-Saclay (France). Maybe in the meantime there are already stronger magnets installed. The race is going on and as is known other similar projects, especially in the U.S. and South Korea, are under development.
Low field MRI systems are permanent MRI systems, ultra-low- fieldMRI (uMRI) systems have a much lower weight, which can be used in some special settings, such as real-time image monitoring on bedside or in the ambulance for human brain disease (stroke). To make the uMRI system more compact, lighter, and smaller even a hybrid method for designing a uMRI electromagnet is already developed. (publication from Pan Guo, School of Physics and Electronic Engineering, Chongqing Normal University, Chongqing401331, China)
Electromagnets are an evenly appropriate alternative for low-field MRI, usually providing better field homogeneity than permanent magnets. In terms of flexibility, they can be turned on and off if wanted. Electromagnets operating at low field are especially interesting as requirements on power and cooling should be drastically reduced. Electromagnets are used as pre-polarising magnets but experimental setups with permanent magnet also exist. A broadening of the spectrum from pre-polarised MRI systems using ultra-sensitive magnetic sensors up to permanent and resistive magnets in compact designs are described in the literature.
Although the high field MR systems are bought since the beginning: a High field MRI system at 1.5 T in the mid-1980s has very little in common with the same system in 2020: field strength alone is not what makes performance.
Many justifications are debated nowadays which bring the low field MRI systems into considerations: simpler cooling strategies, permanent magnets, less MRI artefacts, an increase in T1 contrast, costs reduction, lower sensitivity to magnetic susceptibility effects and enhanced contrast especially in the mid-field systems. Also, the new RF sequences, k-space trajectories and speeding up methods are changing the playground.
One of the main MR manufacturers (Siemens Healthineers, Erlangen, Germany) even launched in 2020 a 0.55 T (23.4 MHz) for a variety of applications.
The natural trend is also to better control medical expenses worldwide and reconsidering low-field MRI that could lead to the democratisation of dedicated, point-of-care devices and decrease high-field clinical MRI systems. And is it likely that the total number of machines worldwide will grow so to follow the increasing demand considering their overall cost? Half of the world population is underserved regarding MR diagnosis, the medical community indeed strives for more universal and accessible systems.
weightbearing MRI system
Hyperfine’s Swoop™ Portable MR imaging system
But back again to the ultra-high field MR at the other side of the broad spectrum of MRI systems.
In 2020 GE Healthcare got an FDA (USA) approval for their 7 Tesla, not only for research but also for clinical imaging. Extraordinary detail and resolution for anatomy, function, metabolism and microvasculature throughout neurological and MSK diseases.
The question is coming up: how to determine choosing the right MRI system for your needs, an improvement in diagnostic capabilities or to realise quantity and quality improvements and not forgetting questions about capital planning.
Future continuous education is needed?
Do we need in the future still radiological specialists? We go from morphology to quantitative MRI. Do we still have to understand the MR system in detail if remote scanning, artificial Intelligence and “one button push” is routinely used and protocols and RF sequences are pre-fabricated? Even when Synthetic Imaging, computed MRI will be regularly used and only one 3D RF sequence will be needed? Will patient-positioning still be important?
At the moment it’s not the MRI system but the MRI technologist who plays a crucial role in the resulting examination quality, as preparation including communication with the patient and protocol set up and fine tuning are crucial to achieve high quality, high resolution images. We will see, but now we still like to fine-tune protocols, shorten the acquisition times and get the maximum out of it! But we should be aware and consider the change in the work of radiologists and maybe even the disappearance of radiological technologists who know the morphology and technology and know what they are doing, not becoming “button pushers”.
As Steve Jobs is saying: “You’ve to start with the customer experience (user satisfaction) and work back toward technology, not the other way around”.
To my opinion there is enough time to perform MR examinations with advanced techniques, thanks to the shortened scans like Compressed Sensing and Multi-Band for all routine RF sequences. It is just important to choose the additional advanced RF sequence between those with the most impact, the less acquisition time and simple post-processing. The workflow and follow-up cases can often be improved to increase the report quality and spare time.
I have always been very interested in the technical part of MRI: figuring out how everything is implemented in order to be able to fine-tune protocols. Of course the clinical part is mandatory too, so I include several clinical applications for the different RF pulse sequences.
I will also describe the importance of the contrast settings and other relevant parameters.
This document should not be construed to represent a definitive interpretation of the regulatory statutes regarding MRI and MRS, and the reader should be aware that regulations might change and render possible out-of-date information specified herein.
Although every attempt has been made to verify the information contained in this e-module, the author cannot guarantee its 100% accuracy. Each effort is made by the editorial board to see that no inaccurate or misleading data, opinion or statements occur. EMRIC sarl cannot accept responsibility for the completeness. This document may not be distributed or re-posted without the express written permission of EMRIC sarl.
I know not everything can be covered, but at least I would like to give an overview of MR knowledge, trying to stay updated, and also putting together what I picked up in the last 36 years. I still like to encourage everybody to do the same!
If you find an aspect of MRI and MRS that I have not covered, do let me know and I will resolve or add it.
1.02 Introduction: What is an RF Pulse Sequence?
A pulse sequence is a pre-selected set of defined RF pulses (non-selective, selective, profiled) and gradient wave-forms, usually repeated many times during a scan, wherein the time interval between pulses and the amplitude and shape of the gradient wave-forms will control MR signal reception and affect the characteristics of the MR images. Pulse sequences are computer programs that control all hardware aspects of the MRI measurement process.
Pulse sequences are commonly listed as the repetition time (TR), the echo time (TE), and if using Inversion Recovery, the inversion time (TI) with all times given in milliseconds, and in case of a Gradient Echo RF sequence, the flip angle. For example, 3000/25/1000 would indicate an Inversion Recovery pulse sequence with TR of 3000msec, TE of 25msec, and TI of 1000msec.
Specific pulse sequence weightings are dependent on the field strength, the manufacturer and the pathology.
So it is a time description of RF, gradients and data acquisition. (see eventually basic MRI in books or internet)
It is possible to make all kinds of RF pulse sequence combinations, which all give different contrast and have different side-effects.
Become familiar with all the RF pulse sequences on your MR system. On most systems you can make a choice between about twenty T1 weighted RF sequences!
Chapter 2 K-space trajectories: essential basics for understanding the e-module
2.01 K-Space (data matrix) and K-space Trajectories
K space is an abstract concept and refers to a data matrix containing the raw MRI data. So it represents the matrix where the MR data will be stored previously to a Fourier transformation to obtain the desired image. It is essential to spatial encoding in MRI where the RF of the magnetization is a function of the local magnetic field strength and a field gradient will be put over an object and will result in different frequencies along the direction.
Filling the k-space produces a full resolution image. Understanding the terms frequency and phase are essential to understand the principles of image formation as the use of the gradients during and in between the RF pulse sequences.
Each row of k-space contains the raw data established under a specific phase encoding gradient, where the order in which the rows (lines) are recorded depends on the imaging sequence used; once all of k-space lines have been assembled, it is Fourier transformed (2D FFT) to obtain the image.
Starting to explain the RF sequences and the main side-effects you really have to understand the basics of the k-space: the k-space trajectories, the possible, available ways how to go through k-space specifically with the T(F)SE RF sequences because it can influence a lot the contrast and even the sharpness depending on the way the RF sequence is implemented on the MRI systems of the different vendors.
Want to learn more? There is a huge number of movies on YouTube where k-space is explained in detail. Enjoy!
Several ways to go through k-space: k-space trajectories
Some notes to start with:
- Increasing the k-values (number of echoes or profiles in the k-space) represents a higher resolution.
- Information about details are in the periphery of the K-space (high spatial frequencies).
- The central part contains contrast-information (low spatial frequencies).
- Finer grain sampling can also result in a wider FOV.
Examples of a single-shot Cartesian grid strategy with a rectangular spiral – and a octagonal spiral path.
Courtesy: Donald W. McRobbie (from picture to proton)
Spiral imaging can produce a very complex pattern in the image texture, since this single shot technique moves on a spiral through the k-space, which can be achieved by oscillating gradients with a phase shift of 90° in the x and y directions. This technique requires data interpolation in k-space to bring the measured data onto a Cartesian coordinate system before Fourier transform. This interpolation can produce fake artefacts with the consequence that the image texture is dependent on k-space interpolation and image reconstruction. The spiral scanning shows less flow artefacts and is faster than the Cartesian method but the images are more blurred due to off resonant spins which can come from B0 field in-homogeneity, local susceptibility effects and chemical shifts. To reduce this blurring, water-only excitation can be used or during the pre-scan a fieldmap can be made to determine the frequency offset (two single shots with different TE’s making a phase map by complex subtraction. The correction and results can be put in the spiral gridding algorithm.
To my opinion the Radial scanning will be explored more in the future reducing the sampling. One of the reasons could be that the Cartesian sub-sampling shows back-folding or phase-wrap.
The newest RF sequences like STAR VIBE have this already built in and even the (XD)GRASP (Siemens) can produce key images restrictively reconstructed every 2 seconds.
Radial mask with 43 (out of 256) projections, i.e., reduction factor 6. Courtesy: Xiaojing Ye
Example of a 3D radial -k-space sampling used in Ultra-short TE RF sequences. Courtesy: Cheuk F. Chan
The advantage using those RF sequences show also that, specifically with abdominal imaging, high resolution ( measuring all k-space lines) gives physiological motion!
A non-cartesian technique that oversamples the centers of k-space using separate cartesian vanes. Easier correction than with radial or spiral sampling schemes that are vulnerable to phase differences and off-resonance effects. Propeller, Multi-Vane, (Rapid) RADAR, BLADE and JET are using this method.
Radial (elliptical) view ordering of 3D-k-space. Elliptical centric acquires data in order a-b-c-d-e; reverse elliptical centric in order e-d-c-b-a.
A very complex signal and texture situation is present in so-called single shot imaging techniques like echo planar imaging (EPI), where k-space is filled in one shot with multiple gradient echoes. This is achieved by a gradient scheme in which the upper corner of the k-space is reached by a single gradient pulse followed by a series of blips resulting in a rectangular movement through the K-space. This technique is very sensitive to local susceptibility artefacts, resulting in image distortions and strong T2* contrast dependence.
And a very creative k-space filling from MRI questions.com:
Radial trajectories are sensitive to even smallest imperfections in the gradient and sampling hardware. Such imperfections may originate from a wide range of sources and depend on the vendor, the system type and even the specific MRI scanner used.
The radial method (or “separate cartesian vanes” as others call it) is implemented in the Motion Reduction Technique: PROPELLER -MULTI-VANE-BLADE-JET-RADAR. Also in some 3D T(F)GRE methods in one of the phase encoding directions.
An example from different k-space trajectories used on GE MR systems using diverse RF sequences. Those RF sequences will be discussed in later chapters.
The ky-kz segmentation scheme for TRICKS (a), CAPR (b) and DISCO (c) and the corresponding subsampled PSF (Point Spread Function): d-f. Note that the PSF for DISCO (f) shows significantly reduced coherent artefacts around the main lobe and the periphery compared with that TRICKS and CAPR. Comparable sections from a 3D acquisition on a resolution phantom, where the phantom was displaced by 5mm midway through the scan are shown in (g-i). Note that the artefact (arrows) are stronger and more coherent in TRICKS and CAPR (g.h) compared with DISCO(f), agreeing with the PSF simulations. TRICKS stands for Time Resolved Imaging of Contrast KineticS. CAPR stands for Cartesian Acquisition with Projection Reconstruction. DISCO stands for Differential Subsampling with Cartesian Ordening.
Web links regarding k-space trajectories:
- The Basics of MRI, e-book by J.P.Hornak.
- MR Technology Information Portal. Go to Database and search for K-space.
Although many sites are not free accesible anymore as before which also give nice and clear information about k-space trajectories:
- What is k-Space tutorial on the Revise MRI Physics site. See also their k-Space tools.
- MRI physics, an exposition by LeedsCMR.
- Breaking the speed limit in MRI. An article on fast imaging methods by M.S.Cohen et al.
- Echo-planar imaging (EPI) and functional MRI, by M.S.Cohen.
- An Improved Analytical Design for Archimedean Spirals in 2D K-Space, by M.Amann et al.
- MRI on the Web. Go to Principles,K-space .
2.02 Speed up the acquisition times by going faster through the k-space
Several ways of reducing the acquisition times with help of the k-space are:
01. Reducing the acquisition in the y or x direction (axials as originals)
During the reconstruction the pixels can be” interpolated” (zero-padding) getting the same size of pixels (isotropic) but during the acquisition an-isotropic pixels are acquired. The phase encoding steps maybe varied: the high order profiles are not measured. The resolution is reduced and the S/N Ratio is increased. Ringing or truncation artefacts can appear: filters can be applied.
02. Using Half Scan – Half Fourier – Fractional NEX
A little more than half the raw data is sampled. During the reconstruction the other half is reconstructed (copied). This is possible because of the underlying symmetry of the echo. The S/N ratio is reduced. The resolution is kept the same but the sequence is more sensitive to motion (motion during measured lines can be copied).
03 Using Partial- or Half- or asymmetric- or Fractional Echo / matched Bandwidth
A reduction of profiles / lines measured in the frequency direction: a little more than 50%.
So shorter TE’s can be chosen, less susceptibility and less flowartefacts, as smaller FOV and stronger angulations.
04 Reducing the phase- or asymmetric- or rFOV
A non-square FOV can be made in phase- and frequency encoding direction.
The resolution stays the same but the S/N ratio is worse, measuring less lines in the k-space.
05 Acquisitions which rely on dynamic scans to reduce the scan time such as Keyhole, RIGR, DIME, UNFOLD, K-t Blast
An example of Keyhole Imaging:
A user defined number of dynamic scans with reduced matrix can be launched in which a reference scan is first created before, for example a contrast agent is injected. If needed the reference scan can also be the last with a full matrix. Subsequently, during injecting the contrast agent, the recording of this, is followed by the restarting of the dynamic series. The reference scan is created with a full array so that the K-space is fully filled in. For the reference scan a larger number of measurements can be selected to create an image with both good SNR and optimal detail display. Only the contrast-determining phase encodings are made of the dynamic scans, so that only the middle lines of the K-plane are filled in.
Advantage is that the detail view is good by using the strong phase encodings of the reference scan while the contrast display is determined by the short dynamic scans. So, it can be scanned very quickly while maintaining detail. The number of profiles measured during dynamic keyhole scans is expressed in a percentage of the entire array (keyhole percentage). This keyhole percentage is set by the user and can be between 15% – 100%. In general, a percentage of 25% – 30% will be selected. A low keyhole percentage (e.g. 15%) has the advantage a short scan time (which allows the course of the dynamic process to be displayed at short intervals) but reduces contrast display. The use of an RFOV (reduced FOV) affects both the reference scan and keyhole recordings. With a chosen keyhole rate of 30% and an RFOV of 50%, the actual keyhole percentage will be only 15%! It is of the utmost importance that during the entire recording time no movement occurs in the chosen slices. The measurement data obtained in the K-space of the reference scan does not fit correctly into the K-space of the keyhole images. Aliasing is therefore selected in the direction in which the least movement is expected. For example, if a coronal scan of the kidneys is made, the fold-over will have to be towards LR to avoid movement due to breathing. Keyhole is not applied to heart and lungs because too much movement takes place in these areas. Keyhole can be applied in SE and GRE RF sequences. The scanning time reduction at T(F)GRE RF sequences using pre-pulses is less spectacular. Half scan, Half NEX or Half Fourier can be added.
Also, the Time Resolved MRA scans have basically this method like TRAQ, TRICKS, 4D TRAK, TREAT, TWIST and FREEZE FRAME which is mostly called View-sharing.
Example of K-t Blast:
k-t Blast has a spatio-temporal correlation so in space and time.
06 K- space shutter / elliptic centric view ordering in 3D
It makes an elliptical shutter in k-space and skips profiles outside the shutter. S/N ratio and resolution are better. Between 25-80% scan% k-space shutter can be used. It is differently implemented on the MRI systems. (E.g. above 80% normal scan reduction is done on the Philips systems showing a bit more blurring than K-space shutter.)
07 pMRI: Parallel Imaging
It is a multi-fold increase in imaging speed using multi-element / phased array RF coils. Without needing faster-switching gradients and without additional RF power deposited.
There are 2 approaches:
Image based: methods that reconstruct images from each coil element with reduced FOV and then merge the images using knowledge of individual coil sensitivities ((m)SENSE, ASSET, SPEEDER, RAPID), so reconstructed and then adjusted. Data acquired from each coil element goes into reconstruction of whole image – reduces imaging time, reduces SNR, reduces uniformity.
There are 3 parts needed in the SENSE method: a preparation phase which determines sensitivity maps of individual RF coil elements, position of the coil and the signal picked up.Then an acquisition part which measures less k-profiles and the reconstruction part which decodes images using coil element sensitivity maps.
K-Space based: Methods that explicitly calculate missing k-space lines before Fourier transformation of the raw data (SMASH, GRAPPA, ARC and fastSPIRiT) so adjusted and then reconstructed. Although, those parallel imaging techniques are not exactly the same implemented GRAPPA, ARC and fast SPIRiT use auto-calibration lines to find linear weights to make missing k-space so coil sensitivity maps are not needed.
Artefacts with a noise band and aliasing which appears in the center are strongly shown in SENSE /ASSET/ SPEEDER/k-RAPID as in SMASH/GRAPPA.
The parallel Imaging techniques are also evolving: a newer method is e.g. (i)GRASP (Golden-angle Radial Sparse Parallel MRI), in development by different vendors that combines compressed Sensing, a golden-angle radial sampling with parallel Imaging.
The advanced ways of reducing the acquisition times with help of the k-space are:
- More advanced Parallel Imaging Methods with high reduction factors (like ARC and Caipirinha and fastSPIRiT).
- Controlled Aliasing in Parallel Imaging Results in Higher Acceleration).
A 47-year-old woman with metastatic medullary thyroid carcinoma shows a 1cm metastasis in right hepatic lobe. It was rated as score 2 (optimal or excellent visualization) on 5 minutes delayed postcontrast CAIPI-VIBE (left) as well as standard GRAPPA-VIBE (right) images by majority of readers. The VIBE RF sequences is a 3D T1w GRE: Volumetric Interpolated Breath hold Examination
- Reduced or small FOV Imaging: aliasing artefacts disappear without any phase oversampling. There are various reduced FOV approaches implemented on different systems and on high field systems because of B1 inhomogeneity problems and multi slice signal loss. The eldest approach is used in MRS where 2 or 3 perpendicular applied RF pulses are used to excite a row of voxels or a single voxel (e.g. SVS-Single Voxel Spectroscopy). The tissue outside the small FOV can also be suppressed by slabs to null signal.
Multi slabs used
The advanced method used by the manufacturers to apply a small FOV (like FOCUS, Hyper Cube, ZOOMit and iZOOM) uses a specific 2D RF excitation pulse which is spatially selective in the slice select and phase encoding directions. On most MR systems this method can be combined with nearly all the RF sequences.
Reduced FOV MRI with two-dimensional spatially selective RF excitation and UNFOLD showing from left to right: full FOV, half FOV and quarter FOV. The phase encoding is in the vertical direction
- CS-MRI (Compressed SENSING or Hyper SENSE or Speeding Up). Three main parts of CS-MRI are random sub-sampling, transform sparsity and non-linear iterative reconstruction. Images are “SENSED” and “COMPRESSED”: it subsamples (randomly chosen sets of pixels) a signal. Less echoes sampled from the high spatial frequency are of k-space resulting in some image blurring. It is a smart reconstruction and processing outline. The Siemens CS flowchart shows a step-by-step visualisation of the CS measurement and reconstruction process.
Courtesy: Siemens Magnetom Flash (66) 3/2016 Matthias Blasche
Patients with limited breath-hold ability or are unable to follow breathing commands, e.g., patients with dementia, hearing impairment, children or multi morbid patients, can now undergo e.g. high-resolution dynamic abdominal imaging with free-breathing. It is also very useful in cardiac imaging: stress perfusion, Q flow, T1 mapping and adjust cine imaging to the needs of the patient. (e,g. BH 11×4 sec or 2×15 sec)
Courtesy: Journal of Cardiovascular Magnetic Resonance – BioMed Central
Compressed SENSING (Parallel Imaging)
Multi-Band (Hyper Band) or SMS (Single Multi Slice Imaging): Simultaneous Excitation of multiple slices with phase shift between slices. It is not really a k-space acceleration trajectory but, because more slices can be excited simultaneously, the acquisition time can be shortened massively.
The advantage is an acceleration in data acquisition that is equal to the number of simultaneously excited slices. Different to in‐plane parallel imaging this can have only a marginal intrinsic S/N Ratio (Signal‐to‐Noise Ratio) consequence, and the full acceleration is possible at fixed TE’s, as is required for many EPI applications. Additionally, some implementations SMS techniques can reduce RF power deposition.
This technique has reduced the acquisition time of multi-slice EPI by more than tenfold and has lately become very useful for functional MRI studies improving the statistical definition of neuronal networks and DTI (Diffusion Tensor Imaging) for improving the ability to visualise structural connections in the human brain. So now DTI can be done in a routine exam and fMRI with a high temporal resolution!
The combination of Compressed SENSING with parallel imaging has the advantage of improved image quality, however it comes at a cost. These algorithms involve substantially more computation than direct or iterative linear reconstructions. But the processors are getting so powerful nowadays and even more advanced CS techniques are developped and implemented!
If you want to go deeper into the parallel Imaging methods (m)SENSE, ASSET, ARC, RAPID, SPEEDER- and SMASH (Grappa) as basic methods, or even looking into the advanced speeding up methods to reduce the acquisition times: dive again onto internet: they are all available on the latest MR systems. And the advanced acceleration techniques are also discussed in the Neuro and MSK / Orthopedic e-module.
2.03 Technical Review
LEARNING – EVALUATION: K-space trajectories
After this introduction of k-space and at the end of the module you should be able:
- To understand the principle aspects of k-space.
- To have an in depth knowledge of the conventional ways going though k-space.
- To understand the indication and techniques of the conventional k-space trajectories.
- To understand the importance of the different k-space trajectories related to the contrast.
- To describe various applications for different k-space trajectories.
- To apply techniques to reduce the acquisition time with help of k-space reduction techniques.
- At the end of this e-module to have the skills to design imaging protocols for MRI exams with a certain k-space trajectory choosing the best method and contrast.
There is a k-space tutorial as an MRI educationtool for a good understanding of the k-space and where you can add filters and artefacts, described in the Biomedical Imaging and Intervention Journal 4(1):e15DOI: 10.2349/biij.4.1.e15 : https://www.researchgate.net/publication/51166907_kSpace_tutorial_An_MRI_educational_tool_for_a_better_understanding_of_k-space
I suggest you look on your MRI system which K-space trajectories are available, which can be combined.
There are also many online k-space tutorials and k-space questions available.
Before you go to the next part:
Answer the questions and choose a statement so you can jump to the next part!