3.2 Clinical Indications for MRI of the Brain

The common clinical indications for brain MRI include:

·       Multiple Sclerosis (MS)

·       Primary tumour assessment and /or metastatic disease

·       AIDS (toxoplasmolisis)

·       Infarction (cerebral vascular accident (CVA) versus transient ischemic attack (TIA)

·       Haemorrhage

·       Hearing loss

·       Visual disturbances

·       Infection

·       Trauma

·      Unexplained neurological symptoms or deficit.

                                          (Westbrook, 1999)

3.3    Image interpretation of brain MRI

The following is a summary of MRI Tissue characteristics of brain structures adapted from Woodward (2001).

Table 1 Tissue characteristics on MRI

Being given the normal appearance of brain structures on MRI we can now move a step further and try to identify any abnormalities in the underlying images.  This may be performed by the process of patterned recognition.  Woodward (2001) presents three rules of thumb, which may help with the search of abnormal areas:

·        Are the right and left halves of the brain symmetrical?

Both halves of the brain should appear symmetrical, if not there is a good chance it’s an abnormality

·        Are there areas that exhibit any signs of mass effect?

Pathological processes as tumours, usually take more space than the original structure, either by destroying it or pushing it aside.  The ventricles should be intact and symmetrical and the sulci and gyri must be clear, without swelling and distortion. 

·        When looking at T2-weighted images are there any bright signal intensities other than the CSF spaces and venous structures?

Most pathological processes involve the presence of oedema, which appears bright on T2-weighted images.

The table below will help in the recognition of normal anatomy and pathology based on MRI contrast characteristics devised Westbrook (2001).

Table 2 Pathology characteristics on MRI

3.4    Image optimisation in MRI of the brain

3.4.1        Equipment

For the above indications a dedicated head coil (quadrature or phased array) is essential in routine MRI if the brain.  Most head coils are now equipped with mirrors or prisms allowing vision of the exterior.  Pads and straps are usually essential for patient immobilisation and earplugs are supplied to prevent any damage to the ears from the gradient noises.  For centers performing Echo-Planar Imaging (EPI), diffusion and perfusion imaging high performance gradients are required (Westbrook, 1999).

3.4.2 Image quality factors

The most common factors, which should be taken into consideration when discussing MR image quality, include signal-to-noise (SNR) Raito, image contrast and spatial resolution (Edelman, Hesselink and Zlatkin, 1996). 

The relationship between image data and noise is expressed as SNR.  An image having a high SNR is usually sharp and appears less grainy.  It is therefore recommended to,

“Maximise the SNR to an acceptable level for making a diagnosis without increasing imaging time to a length that is unacceptable to the patient.”

(Edelman et al, 1996 p 61)

Many factors have an effect on the SNR in MR imaging.  A higher magnetic field strength will produce a higher SNR, especially when motion artefacts are suppressed.  Low bandwidth pulse sequences yield images with a high SNR when compared to high bandwidth sequences. The uses of spin echo pulse sequences increases the SNR when compared to gradient echo (GRE) sequences, due to the 90 flip angles that are used were the resultant echo has greater signal amplitude.  As the Number of signal averages (NEX) increases so does the SNR however, this also increases the scan times.  Coherent noise (e.g. respiratory or pulsatile movements) and incoherent noise (e.g. RF leak) both reduce the SNR.  Moreover, receiver coil sensitivity governed by its design and receiver coil distance from the tissue of interest will also affect the SNR resolution  (Edelman et al, 1996; Mitchell, 1999; Westbrook, 1993).

The spatial resolution of an image determines the viewer’s ability two points as separate and distinct and is determined by voxel size and SNR.  If the spatial resolution is increased, the SNR decreases in direct proportion to the volume of the voxel.  Therefore, a coarse matrix, a large field of view (FOV) and a selection of thick slices will improve the SNR but will unfortunately yield a poor spatial resolution  (Edelman et al, 1996; Mitchell, 1999; Westbrook, 1993).

Tissue contrast refers to the differences in signal intensity from different tissue structures.  Tissue contrast (e.g. T1 or T2 contrast) may be manipulated through the selection of scan parameters and pulse sequences.  Some factors include the repetition time (TR), echo time (TE), matrix, NEX, slice thickness, orientation and location and bandwidth.  The magnetic field strength and the coil also effect image contrast (Edelman et al, 1996).

3.4.3 Routine brain MRI

The SNR and contrast characteristics in the brain are usually good.  A dedicated quadrature head coil is used yielding a high and uniform signal, and good spatial resolution at relatively short scans times.  The routine brain examination at St. Luke’s Hospital includes the following:

·        3-planar T2* FGRE

·        Axial PD & T2 FSE

·        Axial FLAIR irFSE

·        Coronal T1 SE

·        Sagittal T2 FSE

A standard study usually starts with a scout multislice acquisition using a fast gradient echo sequence (FGRE), with T1-weighted contrast (Jager et al, 2001).

This is followed a SE/FSE axial dual echo sequence providing proton density contrast on the first echo and T2 contrast on the second echo.  The first spin echo is generated early by selecting a short TE.  Since only little T2 decay has occurred, T2 contrast between tissues is insignificant in this echo.  The second spin echo is generated much later by using a long TE, whereby T2 contrast is now much greater.  In dual echo sequences the TR is long in order to minimise T1 contrast in different tissues (Westbrook and Kaut, 1993).

Diagram 3 A Spin Echo PD/T2 (dual echo pulse sequence)

SE/FSE sequences are used since they produce relatively good grey/white matter contrast at relatively short scan times.  Since the TR used is of the range 2000-2400ms, CSF signal intensity increases due to its high proton density that may reduce the contrast between CSF and periventricular lesions (e.g. MS).  This explains the use of the dual echo sequence (Westbrook, 1999).

Many centers add a sagittal SE/FSE T2-weighted acquisition in routine brain MRI because they feel that this aids detection of corpus callosum involvement in MS (Jager et al, 2001).  For T2 weighted images a TR of 4000ms or higher may be used with an echo train length (ETL) of 16 or more as opposed to an ETL of 4 in the dual echo acquisition.  However, as the ETL increases the possibility of image blurring also increases (Westbrook, 1999).  Westbrook (1999) also suggests the use of a coronal SE/FSE PD/T2 (dual echo) as part of routine brain imaging.

Brown and Semelka (1999) state that a number of innovations have been developed for brain examinations.  These include magnetisation transfer suppression (ideal for background water suppression), diffusion and perfusion weighted images (sensitive in acute stroke) and echo train spin echo (ETSE) fat suppression imaging.  In our MRI unit at St Luke’s Hospital, the FLAIR sequence is used in routine brain imaging.  FLAIR is a variation of the inversion recovery pulse sequence.  In flair, the signal from CSF is nulled, by selecting an inversion time (TI) corresponding to the time of recovery of CSF from the 180-degree pulse to the transverse plane and there is no longitudinal magnetisation present in CSF.  A TI of 200ms is usually required to suppress the CSF signal.  Since the CSF signal is nulled pathologies adjacent to the CSF are seen more clearly e.g. MS  (Westbrook and Kaut, 1993).

3.4.3.1 Fast imaging techniques

There are two ways to speed scan times in MRI.  Primarily, the amount of data collected can be reduced or secondly, to reduce the time span to acquire the number of data points needed.  The former can be achieved by using conjugate symmetry or a rectangular FOV where only a portion of the raw data is sampled while maintaining resolution through zero filling (Woodward, 2001).

Data collection time can be reduced by a number of ways, for example by reducing the

NEX and the matrix size.  However, as stated in section 3.3.2 SNR and /or resolution may be affected.  Reduction in scan time may be achieved by using fast spin echo (FSE) sequences.  Woodward (2001) states that FSE uses a modification of the SE sequence, where multiple data lines per sequence is acquired.

Diagram4 The Fast Spin Echo Pulse sequence

RF=radiofrequency, SS=slice select, PE=phase encoding, FE=frequency encoding

Technique:

As in routine spin-echo imaging, a 90-degree RF pulse is followed by a single 180 degree refocusing pulse. However, in fast spin echo (FSE) imaging, multiple 180 degree refocusing pulses are performed with multiple resulting spin-echoes. Each echo has a different phase encoding value and is assigned to a different k-space line. As the number of echoes acquired after a 90-degree excitation increases, scan time decreases but signal to noise also decreases and T2 blurring increases. Maximum ETL is limited by true T2 decay.

Typical

TR: >3000 ms (for T2-weighted images)

TE: Effective TE is time when the central k-space views are acquired (which determines image contrast).

ETL: Echo Train Length - generally between 4 and 16

                                                                                                            (MARS, 2002)

As in routine spin-echo imaging, a 90-degree RF pulse is followed by a single 180 degree refocusing pulse. However, in fast spin echo (FSE) imaging, multiple 180-degree refocusing pulses are performed with multiple resulting spin-echoes. Each echo has a different phase encoding value and is assigned to a different k-space line. As the number of echoes acquired after a 90-degree excitation increases, scan time decreases but signal to noise also decreases and T2 blurring increases. Maximum ETL is limited by true T2 decay (General Electric, 2002).  Moreover, Woodward (2001) states that the successive 180-degree echoes in FSE reduce its sensitivity to haemorrhage.  To compensate for this factor, modifications involving the incorporation of GRE in addition to the SE have been proposed to increase the level of magnetic susceptibility effects.  

3.4.4        Advanced Fast Imaging techniques

3.4.4.1 Echo planar imaging (EPI)

  Instead of acquiring spin echoes by using consecutive 180 degree refocusing pulses (as in FSE imaging), the sign of the readout (frequency-encoding) gradient is rapidly alternated and gradient echoes are acquired, resulting in markedly reduced scan times. MS-EPI requires several excitations to cover the 64 to 128 views of k-space, while "single shot" EPI does this with one excitation, but requires faster and stronger gradients. Spin-echo EPI images can also be acquired, by rapidly switching gradients in a similar manner, this time following the 180-degree refocusing pulse.

Diagram 5 Single and Multi Shot Echo Planar Imaging

RF=radio frequency, SS=slice select, PE=phase encoding, FE=frequency encoding

Technique:

Instead of acquiring spin echoes by using consecutive 180 degree refocusing pulses (as in FSE imaging), the sign of the readout (frequency-encoding) gradient is rapidly alternated and gradient echoes are acquired, resulting in markedly reduced scan times. MS-EPI requires several excitations to cover the 64 to 128 views of k-space, while "single shot" EPI does this with one excitation, but requires faster and stronger gradients. Spin-echo EPI images can also be acquired, by rapidly switching gradients in a similar manner, this time following the 180 degree refocusing pulse.

Typical

MS-EPI: # of shots required is k-space views desired / ETL

GRE readout: ETL 4, TE (effective) 5 ms, echo spacing 5 ms

SE readout: ETL 4, TE (effective) 80 ms, echo spacing 5 ms

Typical Parameters:

MS-EPI: # of shots required is k-space views desired / ETL

GRE readout: ETL 4, TE (effective) 5 ms, echo spacing 5 ms

SE readout: ETL 4, TE (effective) 80 ms, echo spacing 5 ms

(MARS, 2002)

Below are some advantages and disadvantages of EPI adapted from Mitchell, 1990):

3.4.5        Functional imaging techniques – Diffusion and Perfusion imaging

The random motion of molecules in fluids is responsible for molecular diffusion.  These motions are however independent from the rotational motions responsible for T1 recovery and T2 decay.  Contrast from molecular diffusion is caused because phase coherence of in vivo water protons undergoing random motion is reduced resulting in MR signal attenuation.  By the application of strong encoding gradients in particular directions, signal attenuation can be exploited to produce valuable diffusion variations.  In order to accomplish diffusion weighed images, the attenuation factor B is calculated and used to produce the diffusion based MR signals.

           B = exp (-bD)

Where b is a factor depending on the gradient intensity and duration and D is the diffusion coefficient (Woodward, 2001).  Therefore, by increasing the b value the amplitude of the diffusion gradient will also increase making the acquisition more sensitive to the diffusion of water in normal brain.  The signal from areas with normal diffusion is low while areas of restricted diffusion appear bright (Westbrook, 1999).

“In this way diffusion weighted MRI (DWI) can be used for the early detection of acute strokes, ischemic lesions and other brain and body abnormalities.”

(Woodward, 2001 p137)

DWI is also very sensitive in the early detection of oedema, ischaemia, cancer and cyst differentiation.  Unfortunately DWI are very sensitive to motion and therefore more prone to artefact production.  Hence, the shortest sequences possible as the Single-shot EPI is recommended at the detriment of the in-plane resolution and SNR (Woodward, 2001).

Westbrook (1999) refers to perfusion as the microcirculation or the delivery of blood to the tissues.  Perfusion imaging refers to the to the measurement of blood volume in these areas.  To measure perfusion either stationary spins need be suppressed or signal intensity from perfusing spins enhanced.  Either employing motion sensitive gradients or injection of contrast agents may achieve these (Westbrook and Kaut, 1993).

In order to maximise the perfusion effect rapid bolus delivery is essential were an injection rate of 2-3mls/s is recommended.  Perfusion images are usually achieved through SE-EPI or GRE-EPI sequences. The latter greatly increases sensitivity to susceptibility effects of the bolus of gd-DTPA as it enters the brain on the first pass.  However, air/tissue interface susceptibility artefacts are much more severe with the GRE-EPI than with the SE-EPI sequence.  The SE-EPI is suitable for brain perfusion imaging, providing temporal resolution of enhancing lesions and indicating activity.  However, since SE-EPI is not so sensitive to Gd-DTPA susceptibility a higher dose may be needed (Westbrook, 1999). 

3.4.6        Artefacts in brain MRI

Many different forms of artefacts exist in MRI.  The main source of artefacts in brain imaging is from flow motion of the carotids and vertebral arteries (Westbrook, 1999).  Because blood flows throughout the body, different spins constantly flow into and out of the imaging planes.  Therefore, as the protons absorb the energy, they are not available to return a signal back into the original location and a flow phenomenon is created (Woodward, 2001).  This artefact may be reduced by sending a spatial pre-saturation pulse is placed inferior to the FOV.  If the FOV is large an inferior pre-saturating pulse is sufficient, as there is no flow coming in the FOV from any other direction.  For small FOVs, superior, inferior, right and left spatial pre-saturation pulses are necessary (Westbrook, 1999).

Gradient Moment Nulling  (GMN), which is a flow compensation technique, minimises artefacts especially in the posterior fossa.  However, it not only increases the signal in the vessels but minimises the TE available.  Therefore, it is reserved for T2 and T2* weighted sequences (Westbrook, 1999).  Another method of minimising the flow motion artefact is through ECG or pulse triggering.  Imaging is synchronised with the cardiac cycle so that blood or CSF motion is nearly identical from one phase encoding step to the next.  Therefore, the ghosting that is generated in the along the phase encoding axis is eliminated.  The major problem with this technique is that it is time-consuming (Bogdan, 1999).  Another method is to swap the phase encoding axis with the frequency-encoding axis in order to remove the ghosting from the region of interest (Westbrook, 1999).

Another source of artefact in brain MRI is patient motion.  Artefacts due to patient motion may be due voluntary and involuntary.  Voluntary motion can be usually reduced by making the patient feel more comfortable and restriction the motion through immobilisation.  Involuntary motion occurs due to respiration, cardiac motion, orbital motion and swallowing.  The latter may be reduced by gating and pre-saturation techniques (Bogdan, 1999).  Westbrook (1999) recommends the use of fast sequences for the uncooperative patient.  However, FSE while faster, may produce more artefacts when compared to SE because one of the central k-space lines is being filled during each TR period.  On the other hand, single shot FSE (SS-FSE) is a good sequence for reducing motion artefacts, where a whole brain scan can take up to 30s using ETL of up to 128.  If some degree of patient motion is still present, SS-EPI may be used which on the other hand is prone to air/tissue magnetic susceptibility artefacts (Westbrook, 1999).

Other artefacts may be present in brain MRI imaging, which may be sequence/protocol, related.  On of these are the aliasing artefacts.  The techniques used for spatial localisation assign a unique frequency and phase too each location within the image.  The acquisition matrix and the desired FOV in the phase encoding and readout gradients determine this.  This artefact occurs when the FOV is smaller than the anatomical region and the frequencies for this tissue exceeds the Nyquiest limit, causing frequency wraparound.  This effect may be eliminated by low-pass filters placed in the receivers passing frequencies below the cut-off frequency and attenuating those above.  Aliasing artefacts can also occur in the phase encoding direction when protons outside the FOV are excited.   When these protons undergo phase changes corresponding to the frequency changes the artefact occurs.  Solutions for this artefact may be by simply selecting a larger field of view or the application of pre-saturation pulse to region outside the FOV (Bogdan, 2001; Brown and Semelka, 1999).

The chemical shift artefact may be also encountered in brain MRI.  This artefact appears as a displacement of the fat signal along the frequency-encoding direction relative to other tissues.  This occurs because the protons in fatty acids have a resonant frequency slightly different from the protons in water molecules.  Visually, the image shows a thin, bright band at one edge of a fat –fluid interface and a dark band at the edge of the opposite side.  The chemical shift artefact may be reduced by either suppressing the fat signal or using the STIR technique (Bogdan, 2001; Brown and Semelka, 1999).

Another artefact, which may occur, is referred to as cross talk.  This occurs when there is contamination by an RF pulse in one slice, which may affect the adjacent slice.  This can reduce the longitudinal magnetisation and thus the decreased signal intensity the resulting image from the contaminated slice.  This effect can be reduced by interleaving the slice acquisitions giving enough time for the contaminated slice to return to the equilibrium value.  Cross-talk results in reduced SNR, particularly for tissues with long T1 (e.g. CSF), since these recover more slowly from cross-excitation (Mitchell, 1999).

Substances with weak magnetic properties (paramagnetic) cause magnetic susceptibility artefacts, resulting from alterations of the local magnetic field within the patient.  This artefact is most prominent in non-spin echo sequences although it may be visualised on T2-weighted spin echo images as dark regions (low signal) at the foci of the paramagnetic materials.  On non-SE sequences, e.g. GRE, these foci bloom and appear larger.  Also, these materials can cause marked distortion of the local magnetic field, resulting in signal dropout and the distortion of adjacent anatomy (Cardoza and Herfkens, 1996).

3.4.7        Contrast usage in Brain MRI

Gd-DTPA is a T1-shortening contrast agent currently in widespread use in MRI departments.  The following list indicates those structures normally enhanced by Gd-DTPA:

·        Dural sinuses -    superior sagittal sinus

transverse sinuses

·        Nasopharynx mucosa

·        Choroid plexus

·        Pituitary gland

·        Infundibulum

·        Cavernous sinuses

(Woodward, 2001).

Contrast studies are usually required for tumour assessment such as meningiomas and neuromas.  Active MS plaques and metastases also enhance, especially after high doses.  At St. Luke’s Hospital, certain radiologists prefer to use a triple dose of Gd-DTPA as they believe that tumour enhancement is improved.  Infectious process, such as abscesses, encephalomyelitis and toxoplasmosis are very susceptible to enhancement.  Contrast studies will also help the early detection of these infections and improve the patient outcome (Carolan and Runge, 1992).  In addition, the meninges also enhance so that infectious tuberculosis, leptomeningeal tumour spread and post-trauma to the meninges can be visualised.  Contrast is very useful in determining the age of infarcts.  Recent infarcts can enhance while maximum enhancement occurs when the blood-brain barrier has been breached.  Old and chronic infarcts do not enhance  (Westbrook, 1999).

SE or 3D incoherent (spoiled) GRE are the sequences of choice after contrast administration.  In 3D SPGR the data may be reformatted in the desired plane or location after the scan is complete.  Perfusion imaging, with a rapidly infused bolus of contrast is useful in measuring the activity level of a lesion (Westbrook, 1999).  Following the administration of Gd-DTPA, the use of magnetisation transfer (MT) may increase the conspicuity of contrast enhancing lesions e.g. active MS plaques (Woodward, 2001).

       Back           Contents            Next