Greg Brown SMRT Royal Adelaide Hospital
Prepared in September 1997 for a general audience familiar with basic MRI principles

Introduction Factors Affecting the Appearance of flowing Blood Important Characteristics of Flow Time of Flight MRA (TOF) TOF Applications Phase Contrast MRA (PCA) When is MRA Appropriate?	Summary of Basic MRA Flow Quantification and Contrast Enhanced MRA Flow Quantification MR Contrast Enhanced MRA (GD-MRA) Contrast Enhanced MRA Technique Suggested Readings  References
Back to Greg's MRI Page	Contact the Author

At first encounter MRA images look like a conventional digital subtraction angiogram (DSA) but the similarity is an illusion. When interpreting the images it's important to remember how they are created, what mechanisms create the image contrast and how these methods can emphasize or hide different vascular disease processes.

MR angiograms are usually processed or reconstructed images derived from a set of slices referred to as the "base images". The base images can be acquired sequentially or concurrently with each method being subject to different artefacts. Image resolution can be high in plane (less than 1 mm.), but lower in the slice thickness direction (1-3 mm.) and as a result views at greater angles to the original slice plane will demonstrate step artefacts.

MRA image contrast is primarily created by flow, rather than displaying the vessel lumen. MRA shows blood as it travels as distinct from the lumen it is travelling in. If there is no flow MRA wont show the vessel. Technical parameters of the MRA sequence greatly affect the sensitivity of the images to flow with different velocities or directions, turbulent flow and vessel size. 

Important Characteristics of Flow

MRA images optimally display areas of constant flow velocity, but there are many situations where the flow within a voxel has non-uniform speed or direction. The diagram opposite illustrates the streamlines of flow in the normal carotid bifurcation showing regions of reversed and crossed flow. In a diseased vessel these patterns are even more complex. Similar loss of streamline flow occurs at all vessel junctions and stenoses, and in regions of mural thrombosis. It results in a loss of signal, due to the loss of phase coherence between spins in the voxel. This effect must be considered when interpreting images or selecting technique. This signal loss, usually only noticeable distal to a stenosis, used to be an obvious characteristic of MRA images. It is now minimized by using small voxels and the shortest possible TE. Signal loss from disorganized flow is most noticeable in TOF imaging but also affects the PCA images.

Time Of Flight MRA (TOF)

TOF is the most widely used MRA technique. It provides robust images for a range of clinical questions. TOF use short TR sequences acquired as slices perpendicular to the direction of blood flow. The short TR period saturates stationary tissue driving its MR signal down. By contrast the flow related enhancement (FRE) effect yields a high signal from blood moving into the slice making the vessels stand out bright against a grey background. Maximum flow signal is achieved when a totally new column of blood enters the slice every TR period.

This relationship makes the technique sensitive to flow velocity. TR and slice thickness must be appropriate to the flow velocities expected and even small changes in slice thickness can affect the performance of the MRA sequence. Saturation bands can be placed adjacent to the slices to selectively destroy MR signal from blood flowing in from one side of the slice. This allows the technique to be flow direction sensitive and useful for creating arteriograms or venograms. TOF MRA is acquired with sequential 2DFT slices, 3DFT slabs or multiple overlapping thin 3D FT slabs (MOTSA) depending on the coverage required and the range of flow velocities under examination. The images can be processed by Maximum Intensity Projection (MIP) to interactively create projection angiograms is a range of views. For complete interpretation the base slices should also be reviewed individually and with multi-planar reconstruction (MPR) software.

Applications of TOF MRA

• Cerebral arteriograms
• ? Aneurysm, physical basis of stroke
• Neck angiography
• Skull base lesions
• Atherosclerosis
• post-traumatic vertebral artery dissection & occlusions
• Renal arteriography for transplant donors
• Pelvic and peripheral angiography

• Complex flow or turbulence causes signal loss (Improved by reducing TE)
• Inefficient coverage of long vessel segments
• Short T1 tissues (fat, blood, Gd enhanced regions) can appear the same as flow, obscuring vessels
• Patient motion significantly degrades image quality

Phase Contrast Angiography (PCA)

During an MRI sequence, spatially varying or gradient magnetic fields are applied to spatially localize signal. The method assumes that everything remains stationary. If signal producing tissue (like blood) moves significantly during the sequence the spatial phase encoding is incorrect and a ghost image of the object will be seen mis-mapped in the phase encoded direction. The amount of phase error depends on the velocity of the movement and the strength and period of the applied gradient. In a motion compensated sequence these artefacts are avoided.

The PCA method acquires two sequences; one with and one without motion compensation. By comparing the phase of signals from each location in the two sequences the exact amount of motion induced phase change can be determined so we have a map where pixel brightness is proportional to spatial velocity. Regions that are stationary remain black while moving regions are represented as grey to white.

In its basic form the PCA sequence only shows a signal where a voxel contains uniform velocity in a specific direction. By acquiring 3 sequences each with motion compensation in the three orthogonal directions a complete map of flow regardless of direction is obtained as well as the motion.

Phase change is measured on a circular scale, 0,1,2…358,359,0,1 degrees. PCA sequences are designed so different flow velocities display the maximum 360 degrees of phase shift. This factor is called the velocity Encoding factor or VENC. For best results the VENC is chosen to be slightly more than the maximum expected flow velocity in a region. If a higher velocity is encountered it will be represented as a darker pixel. This is known as velocity aliasing.

2D PCA

A thick slice 3 direction PCA can generate a single projection angiogram in about 1.5 minutes. The method is commonly used to rapidly determine cerebral vein patency, identify venous angiomas and simple AVM, test the appropriate VENC before running a long 3DFT PCA, or provide vascular scout images to accurately position other MRA sequences.

3D PCA

A 3DFT technique is used to partition a thick slab PCA sequence into thin slices which can be processed in MIP and MPR programmes to create multiple projection angiograms. A 3D PCA takes about twice as long to perform as a comparable TOF MRA but is affected by different artefacts and much more sensitive to flow at lower velocities.

2D Cine PCA

A cardiac triggered version of the 2D PCA sequence provides a number of views of the flow at different stages of the cardiac cycle. It is particularly used for pulsatile flow of blood and CSF. The technique uses long sequences and relies on a regular heart rate.

PCA Applications

• Cerebral Veins
• Lesions with flow in various directions
• - AVM
• - Varicosities
• Vascular localiser images
• Hand and foot vessels
• Vessels associated with bleeding lesions
• - Ruptured aneurysms
• - Cavernous haemangioma
• - Bleeding AVM

When is MRA Appropriate?

• Whenever MRA can provide sufficient clinical information to direct treatment
• Where the risks of XRA don't justify the benefits
• Serious contrast reactions
• During pregnancy
• Low suspicion of significant disease
• Investigation of an otherwise healthy individual (healthy donor work-up)
• To augment sectional imaging creating a comprehensive examination

Summary

With appropriate technique, conventional MRA offers accurate vascular information with no risk to the patient in a plainly understandable format. It can provide quick effective display of cerebral vessels, neck vessels beyond the reach of ultrasound, and the peripheral circulation. Its ability to display vessels is limited by spatial resolution particularly the slice thickness. Availability of MR scanners, the speed and capability of current hardware, and acceptance limit its application by clinicians. MRA remains an exciting technique because the latter points are changing rapidly.

MRA Flow Quantification and Contrast Enhanced MRA

Each of these techniques can be used to extend the role of MRA beyond that available with conventional TOF and PCA. Their current implementation as practical clinical tools stem from being able to complete the scan sequences within a single breath-hold period. This is possible because of the high performance gradient systems and phased array coils that are only recently achieving broad distribution.

Flow quantification by MR using breath hold sequences offers the rapid collection of functional vascular information and should alter the role of MRA as significantly as the addition of Doppler techniques did to vascular ultrasound. CE-MRA is being heralded as the technique that will handle the previously marginal task of abdominal MRA, positioning MRA to seriously challenge DSA as the primary vascular imaging modality. 

Flow Quantification (MR-FQ)

The addition of a quantitative analysis of flow rates and time varying blood velocity is used to enhance the value of the MRA examination. MR FQ uses a cardiac triggered 2DFT phase contrast technique. The images can be examined and regions of interest placed over target vessels to generate velocity/time graphs. If the ROI covers the entire vessel, peak and mean flow values can be determined accurately.

Fine temporal resolution is essential to ensure accurate representation of highly pulsatile flow. Conventional MR-FQ sequence times of three minutes make the method susceptible to errors based on heart rate variation and respiratory motion, however they can achieve temporal resolutions of 20-50 mSec. Breath hold sequences offer good results but the temporal resolution may fall to 150 mSec. This can be countered by repeated scans with variable ECG trigger delays as long as the patient can breath hold consistently.
 
 

Effects of Sampling Rate on Curve Representation

Measures of mean flow volumes are most sensitive to inaccuracies in the time/velocity curve as this figure is calculated from the integral of the curve. (area under the curve). The mean flow figures are used to establish if a stenosis actually reduces tissue perfusion (haemodynamic effect). Values can be obtained proximal and distal to the stenosis, or by comparing down stream values to known standards.

• Assessment of valvular regurgitation
• Relative flow measurement in aneurysmal lumen
• Quantifying haemodynamic affects of stenosis or coarctation
• Confirming flow direction

Contrast Enhanced MRA (CE- MRA, Gd- MRA)

CE-MRA is almost directly analogous to IV-DSA and IV-CTA. An intra-venous injection of a short dense bolus of a gadolinium based contrast agent is imaged on the first pass through the arterial system. Sequential scans are added to acquire early and late venous images. Further delayed sequences will record an MR Urogram much like IVP delayed images if required. CE-MRA images are essentially a record of the vessel lumen. Some flow mechanisms affect the contrast of these images, but the sequence parameters are chosen to minimize them. The gadolinium is being used as a blood pool agent to be imaged as it transits the region of interest, although it is not the actual contrast agent that is visible on the images. In high concentrations, the Gd radically shortens the T1 of the adjacent blood. The extremely short TR values (3-7 mSec) ensure saturation of almost all stationary tissues and a high sensitivity to the short T1 gadolinium enhanced blood. Fat signal can remain in the images but these are removed by subtraction. The short TE maximizes signal and minimizes pulsatile ghosts and signal loss from regions of chaotic or non-streamline flow.

The CE-MRA method has been used since about 1993 but it has become reliable and more robust with the advent of breath-hold sequences. These sequences rely on the use of extremely fast sequences and high performance gradient and coil systems. For those reasons the CE-MRA sequences have only been commercially available on high end MR systems since late 1996. Many investigators are working to define the technique's clinical role.

The technique described later in this paper is currently in use at Royal Adelaide Hospital MRI Unit. It is closely based on the method suggested in the Siemens MAGNETOM Application Guide.

CE-MRA Advantages

• Rapid large FOV acquisition
• Good vessel lumen display at stenosis
• Not sensitive to flow direction

• May still not display delayed retrograde collateral filling
• Regional signal loss from stents
• Requires good patient co-operation
• High Contrast cost (and not reimbursed in most instances)

CE-MRA Applications

• Renal Angiography
• Investigated as a primary imaging tool for renal artery stenosis, as a precursor to stent or balloon angioplasty, and for patients with sensitivity to iodinated contrast or suspected familial renal artery disease where TOF imaging has failed.
• Mesenteric & Portal Angiography
• Pulmonary Arteriography
• An ECG triggered technique yields the best results.
• Peripheral Arteriography
• Typically performed with 3 fields of view and 3 separate injections of 0.2 ml/kg (0.1mmol/kg) of contrast. Triggered TOF is used additionally if run-off vessels are not clearly identified.

CE-MRA Technique

Patient Preparation

• Normal contraindications for MRI apply. Arterial stents are safe to image but will obscure display of the repaired region.
• Patients are fasted for 4-6 hours prior to the exam to minimize bowel motion. Sedation with oral Valium is recommended for nervous or claustrophobic patients. A CP Body phased array coil is used.
• The procedure is explained to the patient and the need for them to take consistent breath holds is explained. The breath hold technique is rehearsed. If the patient can't hold their breath for 23 seconds, the are asked to begin breathing gently when they have to rather than starting with a gasp.
• A 20G Jelco is placed in an antecubital vein connected to a minimum volume extension tube and a leur lock Y-connector.
• NB Use syringes with twist on leur lock fittings to ensure secure seal for the rapid injection. Do not use interlink bungs.

• Coronal 3DFT FISP (fi3d_2b488.ykc) (5/2/25^o) Linear phase ordering. Scan time 26 seconds. FOV 350 mm. Resolution 2.5 x 1.37 x 2.5 mm
• A "mask" scan is performed immediately before the injection run.
• For the injection run the sequence is repeated 3 times with a 10 – 15 second inter-scan delays.
• The patient holds their breath on slight inspiration during the scan sequences.

Scan Timing
Proper co-ordination of the scan sequence and injection is vital to the quality of CE-MRA, particularly when using low Gd doses (0.2-0.3 ml/kg). Most of the image contrast is derived during the acquisition of the central ½ of k-space, (7-21 seconds after scan start), so the contrast bolus must be in the field of view during this period.

Timing Run & Establishing Transit Time:

• A sequence acquiring one image per second for 30 seconds is started at the same time as a 30 ml. bolus of saline doped with 1 ml of contrast.
• The bolus is injected at approximately 3 ml./sec. per second and followed immediately by a 30 ml. bolus of saline at the same rate.
• A time/intensity curve is made for a region in the aorta and the arrival time of the test contrast bolus determined. As the injection period is almost equal to half the scan period, the scan is timed to commence so that the bolus will arrive after ¼ of the scan period. This ensures contrast in the field of view during the period critical for image contrast.
• There are apparent conflicts in the methods for calculating an appropriate scan/injection delay. Some authors suggest that the bolus should arrive half way through the scan period but this ignores the image contrast afforded by data collected in the early half of k-space. Then correct calculation of delay times depends strongly on the loop structures of the scan sequence. If a centric phase ordered sequence is used (i.e. GE Medical Systems), the scan must be started when the bolus arrives. GE SIGNA Smartprep uses a dynamic display of MR signal in the Aorta to allow the radiographer to start the sequence interactively in a manner that would be familiar to GE CT users.

• Best results are achieved when everyone is aware of the critical elements of the technique.
• For the timing run and main contrast run rehearse all instructions and the flow of injection and scanning with the scan radiographer, the person injecting, and the patient.
• Use consistent start instructions, called by the Radiographer operating the scanner.
• It is common for Radiographers to inject. Gd-MRA sites should consider training and certification arrangements to formalize this activity.
• If the procedure fails it can't be repeated for about 12 hours to allow for total excretion of the contrast.

• Use a stopwatch to ensure accurate timing.
• Hand injection at 3 ml./sec is difficult for most people. Pushing down on the barrel of a syringe resting on a flat bench works well. A good quality 3 way tap allows fairly rapid changeover from contrast to saline bolus but a Y-connection is faster if the injector learns to pinch off the line to the non-injecting syringe.
• More consistent results may be achievable using MR compatible programmable power injectors. This device can deliver higher injection rates that may yield quality images with lower contrast doses.

• Subtraction of the contrast series from the mask series greatly improves the quality of most images. Signal from abdominal fat is particularly avoided.
• Bowel peristalsis will produce mis-registration ghosts but this can be minimized by fasting the patient for 4-6 hours.
• If the patient held their breath at inconsistent phases the subtraction sets may produce worse results than the native images.
• Creative subtraction of the subtracted early and late venous phase series can yield pure venous images in some cases.

MRI Physics of Flow and MRA. Evan Fram Book of Abstracts the Gold Coast MRI & CT Conference. 1996

SIGNA Applications Guide Volume 3. P.Turski (editor) GE Medical Systems. 1990

MAGNETOM Applications Guide 1 Techniques Section D2. Siemens Medical Systems 1997

Click Me to Mail to Me If you would like this paper presented at your MRI meeting or you have comments or suggestions, please contact the author by e-mail.