British Journal of Radiology (2006) 79, 432-436
© 2006 British Institute of Radiology
doi: 10.1259/bjr/76396327
Assessment of environmental disturbances to the static magnetic field in magnetic resonance installations
M A Schmidt, PhD
Department of Medical Physics, St George's Hospital, Blackshaw Road, London SW17 0QT, UK
 |
Abstract
|
|---|
The static magnetic field of MRI scanners can be affected by environmental factors. Magnetic resonance spectroscopy and functional imaging with single-shot echo-planar imaging (EPI) are particularly vulnerable to the movement of lifts, vehicles, trains and other large metallic masses in the vicinity. This work investigates the sensitivity of two different imaging techniques to assess disturbances of the static magnetic field: (i) phase changes in gradient-echo images of a uniform test object; and (ii) image displacement along the phase encoding direction in single-shot EPI images. For the latter a hexane sample was used, and the separation between CH2 and CH3 signals was taken as a reference. Both techniques were evaluated in a site known to be free of any significant environmental disturbances and validated by inducing a magnetic field disturbance. Both techniques provide valuable information in acceptance tests, allowing MRI users to evaluate and manage the environmental conditions surrounding a scanner. The single-shot EPI technique was found to be highly sensitive, being expected to detect magnetic field fluctuations down to 0.005 parts per million (ppm). The phase images method was found to be less sensitive (0.02 ppm) but is more easily available. The single-shot EPI technique was used in acceptance tests and environmental disturbances to the magnetic field of the order of 0.04 ppm were measured at the isocentre on two separate occasions.
|
Introduction
|
|---|
MRI scanners are known to be affected by external environmental factors. Examples are movement of lifts, vehicles, trains and other large metallic objects, often having some ferromagnetic content, which disturb the static magnetic field surrounding the magnet isocentre. Functional imaging and other techniques based on echo-planar imaging (EPI) are particularly vulnerable: Durand et al [1] have shown that changes in currents associated with a rail line caused significant deterioration on functional brain studies at 1.5 T. Peaks in magnetic resonance spectra can also suffer considerable broadening, reducing magnetic resonance spectroscopy (MRS) sensitivity.
MRI equipment suppliers advise on site planning and on minimum distances between the magnet isocentre and various sources of disturbance to the magnetic field. Some MRI scanners are placed relatively close to possible sources of disturbance, and checking for any environmental effects then becomes a necessary part of the acceptance test. It may also be necessary to verify the stability of the magnetic field to aid the management of the environment surrounding an MRI system throughout its life-span.
At any given point within the magnet bore, a transient environmental disturbance to the static magnetic field will be perceived as a fluctuation of the Larmor frequency. A second effect is a transient reduction of the field homogeneity within a given volume. MRI scanners always adjust the central frequency prior to imaging, but the timescale of this automatic measurement is invariably slow, not allowing the user to measure a transient magnetic field fluctuation reliably. A localized measurement of the magnetic field fluctuation could be made with a small sample, simply by repeatedly acquiring a free induction decay (FID) signal. However, in practice, this is extremely difficult to undertake in acceptance tests, as the required pulse sequence is often unavailable.
This work investigates the sensitivity of widely available imaging techniques to evaluate disturbances of the static magnetic field. Current shimming standards allow the routine acquisition of brain 1H spectra with 2 Hz water line width at 1.5 T, and thus measurement techniques must be sensitive to magnetic field variations at least of the order of 0.01 parts per million (ppm). To be truly useful in acceptance testing, a technique for the measurement of changes to the static magnetic field must not depend on any detailed knowledge of the pulse sequences involved or other information to be provided by the manufacturer. Also, it must not depend on specific software packages, which may not be available. It is also desirable to produce an objective record of any transient changes in the static magnetic field intensity. The latter is particularly relevant for disturbances which are not very frequent, but are expected to be significant (an infrequent train service in the vicinity, for example).
|
Methods
|
|---|
Two separate imaging techniques were considered for the quantification of magnetic field disturbances caused by environmental factors, and their sensitivity was evaluated. This work was undertaken at the Radiology Department, St George's Hospital (GE Signa/Echo-Speed gradients, Milwaukee, WI). A 1.5 T MRI scanner has been in operation on this site for 10 years and the site is known to be free of any significant disturbances to the static magnetic field due to the surrounding environment.
Phase images method
Changes in the magnetic field value are demonstrated as phase changes on gradient-echo images, since the term 
.TE.b(t) is added to the image phase in the presence of magnetic field fluctuations ( is the gyromagnetic ratio, TE is the echo-time and b(t) is an unknown magnetic field fluctuation). This technique can be used in any scanner, but some of the main manufacturers require a research agreement or a service password to produce phase, real and imaginary images. Manufacturers also use different approaches for the scaling of the phase images, sometimes using a threshold from the magnitude images to avoid displaying the random background phase. When the phase scaling for a given scanner is not known, the phase must be calculated from real and imaginary images. Another solution to eliminate any ambiguity is to impose a known shift to the central frequency, as this will cause a known phase change, and acquire two images (before and after shift) for reference.
A self-loaded uniform cylindrical test object with 19 cm diameter was employed (NiCl solution, 1.66 g l–1, pH 4). A relatively short T1 was preferred to maximize the signal-to-noise ratio (SNR) in rapid imaging techniques. The phase change associated with field fluctuations can be maximized by increasing TE, but this slows down the measurement and leads to poorer SNR. As a compromise, a series of 100 fast spoiled gradient-echo images were acquired with the following parameters: TE = 3.3 ms (shortest TE for a symmetrical echo), repetition time (TR) = 7.3 ms, 24 cm field of view (FOV), 10 mm slice thickness, 256x128 data acquisition matrix and ±32.25 kHz receiver bandwidth. These parameters produce a 2
phase shift for a 4.4 ppm shift on the Larmor frequency.
The first phase image acquired was taken as a reference image, and all subsequent phase images were subtracted from the reference image. A region of interest (ROI) was chosen at the isocentre and the ROI mean phase value was computed over the resulting series of subtracted phase images. The range of ROI phase values was used to estimate the sensitivity of the method.
Single-shot echo-planar imaging (EPI) method
Magnetic field fluctuations cause image shifts along the phase encoding (PE) direction in a series of single-shot spin-echo EPI images. The magnitude of the image shifts depends on the PE bandwidth and, as a consequence, on the interecho delay. However, details of the EPI pulse sequence may not be known at the time the acceptance test is performed, since manufacturers differ on the amount of pulse sequence information provided as standard. Again, a known shift in the central frequency will produce a reference displacement, which can be used to calibrate any other images where a displacement along the PE direction occurs. Another solution is to use a sample which contains more than one chemical shift, and use the known separation of the two peaks as a reference. The use of the single-shot EPI pulse sequence is widespread, but it is not necessarily available in every scanner as it is usually sold as a separate software package.
An 8 mm diameter test tube containing hexane (C6H14) was employed, and the known chemical shift difference between the CH2 and CH3 peaks (0.39 ppm) was used as a reference. Both peaks have similar intensity and are close enough to produce two clear images if the central frequency is chosen as the midpoint between the two peaks. A series of 100 single-shot spin-echo EPI images was acquired at the rate of one image per second with the following parameters: TE = 100 ms, bandwidth ±31.25 kHz, slice thickness 10 mm, FOV 24 cmx24 cm, producing a 256x256 image using a half-Fourier acquisition. These parameters were chosen to make the technique as sensitive as possible by increasing the PE bandwidth. The two locations of the sample tube in the images were determined by calculating their centre of gravity, after thresholding. The distance between the two images corresponding to the CH2 and CH3 peaks was also calculated to provide a reference.
Both techniques described above can generate at least one image per second, as this rate was considered the minimum requirement to monitor magnetic field disturbances associated with traffic, trains and lift movement. Images can be acquired at a higher rate by reducing the data matrix size along the PE direction. When it was possible to control the potential source of magnetic field disturbance, this control was exercised (by controlling lift movement, for example). The images were processed with in-house software written in IDL (IDL 6.1; Research Systems Inc., Boulder, CO). Both techniques were validated prior to use in acceptance testing by bringing a strongly paramagnetic solution of dimeglumine gadopentetate (concentrated Magnevist, 0.5 mol l–1; Schering, Berlin, Germany) closer to the measurement point and then removing it, creating a magnetic field disturbance.
After validation of the imaging techniques, the stability of the magnetic field was investigated at two other 1.5 T MRI scanner sites as a part of acceptance testing in new installations (Philips Intera/Explorer gradients, Eindhoven, Netherlands and Siemens Symphony/Sprint gradients, Erlangen, Germany).
|
Results
|
|---|
Sensitivity assessment
Phase images method
Figure 1a
shows the first phase image of the test-object, taken as a reference image, and Figure 1b shows the difference between one of the phase images in the series and the reference image. The maximum phase change for any pixel over the object within the series of 100 images was under 0.25 ppm. Considering the average phase over the small central ROI indicated (Figure 1b), the range of variation was up to 0.02 ppm (0.003 ppm standard deviation). This method is therefore not expected to be sensitive to field fluctuations below 0.02 ppm. When a strongly paramagnetic solution is brought closer to the test object, the ROI phase value changes as shown in Figure 1c.
View larger version (70K):
[in this window]
[in a new window]
|
Figure 1. (a) Reference phase image and (b) subtracted phase image of cylindrical uniform test object, with region of interest (ROI) used for measurements indicated. The variation of the average ROI phase value over a series of subtracted phase images is shown in (c), both in the absence of any magnetic field disturbance, and in the presence of a magnetic field disturbance caused by bringing a strongly paramagnetic solution to the test object vicinity and then removing it.
|
|
EPI method
Figure 2a–g shows sections of several single-shot EPI images of the hexane sample. CH2 and CH3 peaks produce separate images of the sample tube, of similar intensities. The image shift along the phase encoding direction can be seen from Figure 2d–f, as those were acquired when a strongly paramagnetic solution was brought close to the hexane sample, and then removed. In the absence of any magnetic field fluctuations, the position of the geometric centre of each tube image changes by no more than 0.005 ppm (Figure 2h). The standard deviation is as low as 0.0015 ppm. Figure 2i shows how the position of the tube centre changes, for both CH2 and CH3, when the magnetic field is disturbed by moving a tube containing a strongly paramagnetic solution in the vicinity. Despite some progressive blurring associated with the central frequency offset, the distance between the two peaks is kept constant in the range of the measurement. Therefore the EPI method is expected to be sensitive to magnetic field variations of the order of 0.01 ppm, as required for acceptance testing.
View larger version (35K):
[in this window]
[in a new window]
|
Figure 2. (a–g) Sections of several single-shot echo-planar imaging (EPI) images of hexane, showing the image shift along the phase encoding (PE) direction associated with a magnetic field disturbance. (d–f) A paramagnetic solution is brought to the vicinity of the test object, and (g) then removed. In the absence of any magnetic field disturbances any measured image shifts are under 0.005 ppm, as shown in (h) for the image associated with CH3. (i) The separation between the CH3 and CH2 images remains constant even when there is movement of a paramagnetic solution in the vicinity of the isocentre.
|
|
Acceptance testing
The EPI method with the hexane sample was employed in two separate acceptance tests where the site conditions were to be evaluated. Both sites were equipped to perform functional brain studies with single shot-EPI, and one of them was equipped with a spectroscopy package. Magnetic field fluctuations of approximately 0.04 ppm were detected at the magnet isocentre during both tests. At one site the variation was associated with the movement of heavy equipment (image intensifier) along a shielded wall in the corridor adjacent to the MRI scanner room. A sign was placed to prevent that area from being used in a way that could disturb MRI data acquisition. At the other site, the variation was associated with car parking on the pavement close to the external wall on the side of the magnet room. The erection of a barrier was suggested.
|
Discussion and conclusions
|
|---|
The single-shot EPI technique is the most sensitive of the imaging techniques discussed as it is expected to detect magnetic field fluctuations down to 0.005 ppm. However, single-shot EPI may not be available in every system, and it may not always be possible to decrease the PE bandwidth to the same level of sensitivity used in testing on the GE scanner as those parameters are clearly undesirable for clinical applications. The toxicity of hexane must also be mentioned, as it needs to be transported to the site of the acceptance test with due care.
The "phase images" method is less sensitive but more easily available, and deserves further investigation and optimization. It could be made more sensitive with a higher SNR and using longer echo-times. If higher sensitivity is achieved, the method could be used in large uniform test objects (spherical or cylindrical) to detect disturbance to the magnetic field homogeneity over larger volumes.
Both imaging techniques discussed could be further optimized, trading spatial resolution for a higher time resolution. However, standard MRI pulse sequences often impose restrictions to the minimum size of the data acquisition matrix.
The interpretation of the results obtained with both methods assumes that an actual drift of the Larmor frequency is the only cause of the measured phase shifts and offsets. However, these effects can easily be caused by a number of different sources related to the scanner hardware (electronic components warming up, for example). For a scanner operating within specification, hardware-related phase drifts and offsets should be of very small magnitude, if detectable at all. Because of the possibility of simple hardware malfunction, it is essential to associate the measurements obtained with specific environmental sources and to perform reproducible measurements.
Both techniques discussed allow MRI users to evaluate the environmental conditions surrounding an MRI scanner, and can provide the basis for informed discussion on the management of the area surrounding it. The final decision on how much disturbance to the static magnetic field can be tolerated is complex and depends not only on the nature of the work to be performed in a particular scanner, but also on the pattern of the local magnetic field disturbance and its associated frequency. The single-shot EPI method, in particular, was proven to be highly sensitive and is strongly recommended for acceptance testing.
|
Acknowledgments
|
|---|
The author would like to thank Dr Franklyn Howe for very helpful discussions on MR spectroscopy.
Received for publication April 19, 2005.
Revision received October 26, 2005.
Accepted for publication October 26, 2005.
|
References
|
|---|
- Durand E, van de Moortele P-F, Pachot-Clouard M, Le Bihan D. Artifact due to Bo fluctuations in fMRI: correction using the k-space central line. Magn Reson Med 2001;46:198–201.[CrossRef][Medline]
Top
Abstract
Introduction
