Acoustic noise reduction of MRI systems by means of magnetic shielding D. Biloen, N.B. Roozen Philips Applied Technologies, P.O.Box 218/Bldg. SAQ 2121, 56MD Eindhoven, The Netherlands {david.biloen, n.b.roozen}@philips.com, G.B.J. Mulder, H. Boschman Philips Medical Systems, P.O.Box 1/Bldg. QR, 568DA Best, The Netherlands A magnetic resonance imaging system (MRI) produces high acoustic noise levels during scanning. These systems use a strong static magnetic field in the order of a few Tesla. Additionally, these systems use a magnetic gradient field that varies in time. These gradient fields are created by means of a so-called gradient coil, in which electrical are used in the order of a few hundred amperes. Acoustic noise is generated due to the Lorentz forces acting upon the gradient coil and the main magnet outer vacuum chamber. The Lorentz forces on the main magnet outer vacuum chamber are generated by the eddy which are developed in the main magnet outer vacuum chamber as a result of the varying magnetic stray fields of the gradient coil. Combined with the strong magnetic field of the main magnet, large Lorentz forces are developed, resulting in vibrations and acoustic noise. The paper focuses upon the reduction of these eddy current induced vibrations and acoustic noise through magnetic shielding. Measurements on an experimental MRI system show a reduction up to 2 db. 1 Introduction The acoustic noise produced by MRI systems is a cause for growing concern. Price, De Wilde et al. from the Imperial College of Science Technology and Medicine in London have reported extensively on this subject over the past fifteen years. The acoustic noise levels reported by Price and De Wilde varied from 85dB(A) for.2.5t systems to 115dB(A) for 3T systems in the case of fast-pulsed sequences [1]. They also point out that the increasing gradient field levels and slew rates are pushing these values even higher, and that levels up to 13dB(A) have been reported for 3T systems [1]. 2 Magnetic Shielding 2.1 Background The main magnet outer vacuum chamber of the cylindrical 1.5T MRI system described in this paper has a significant contribution in the total noise production, especially in positions outside the central patient bore, such as in front of the scanner. The source of the vibrations and noise in an MRI system is the high amplitude pulsed, or sequences, running in the gradient coil. In Figure 1 part of such a sequence is shown. In practice these sequences can vary significantly in terms of amplitude and frequency content. Typically these sequences are repeated over periods of tens of seconds. Figure 1 Excerpt from a typical gradient sequence. Such sequences are repeated for tens of seconds. There are three main mechanisms by which these lead to noise and vibrations: 1. Lorentz forces are exerted by the strong static magnetic field on the windings, which conduct of several hundreds of amperes. These forces excite the gradient coil and cause it to vibrate. The vibrating gradient coil radiates noise directly. 2. The structural connection between gradient coil and main magnet outer vacuum chamber transmits the gradient coil vibrations to the main magnet outer vacuum chamber. Because the surface of the main magnet is considerably bigger than that of the gradient coil, it
contributes significantly to the total noise production. 3. The alternating running in the gradient coil system cause stray magnetic fields. These stray fields cause eddy to run in the main magnet outer vacuum chamber. The static field exerts Lorentz forces on the eddy, again causing the magnet outer vacuum chamber to vibrate and radiate noise. In Figure 1 we see an overshoot pattern in the gradient current. This pattern has been added deliberately to compensate for the effect of eddy which arise in the magnet outer vacuum chamber. According to Lenz s law, these run in such a way to counter the impinging magnetic field, and thereby also the desired gradient field. Extensive experiments have been performed to investigate the relative importance of these transmission mechanisms in the process of noise generation in an MRI scanner. For the scanner under investigation we came to the insight that the order of importance of these transmission paths is the same as listed above. It was estimated that barring special covers around the main magnet, more than 1dB noise reduction was not attainable by tackling only the first two mechanisms. In the rest of this paper the examination and reduction of the magnetic coupling between gradient coil and main magnet is described. 2.2 Magnetic Shielding Experiments A rather elaborate experimental set-up was created on the basis of a Philips 1.5T Intera MRI system. A special gradient coil was made which was clad in copper sheeting, see Figure 3. Figure 3 Special gradient coil with eddy current screen. Additionally, special interfaces were made on the coil so that it could be fixed to the main magnet outer vacuum chamber with compliant mounts. Lastly, a PVC cylinder was made which fit inside the gradient coil, and made an airtight seal with the main magnet outer vacuum chamber, thus completely encasing the gradient coil. With this system the relative contributions of the different transmission paths could be examined. This paper focuses on the examination of the magnetic coupling between gradient coil and main magnet outer vacuum chamber. First the magnetic field emanating from the bore as the result of y-gradient current was measured using a voice coil. This stray field can cause eddy in the front and back sections of the (electrically conducting) main magnet outer vacuum chamber x y Figure 4 Stray magnetic field from the gradient coil was measured in z-direction using a voice coil along a line in front of the scanner. Figure 2 Standard gradient coil. The results are shown in the form of spectral contours depicting magnetic field strength in terms of Tesla per ampere gradient current as a function of radial position and frequency. On the top the results for the scanner with the shielded coil are shown, whilst the figure on the bottom was measured on a standard system.
Figure 5 Y-Gradient field strength spectral contours, measured with white noise as indicated in Figure 4. Top: contours of special gradient coil with eddy current screen (db ref 1 [T/A]). Middle: contours of standard gradient coil (db ref 1 [T/A]). Bottom: Difference of magnetic stray fields (db). Clearly the z-gradients are shielded more effectively than the y- (and x-) gradient. Note that the shielding due to the eddy current screen is expected to occur at radii equal or larger than the inner radius of the gradient coil, which is approximately 35cm. This is especially well visible in Figure 6 for the Z-gradient, which shows a reduction of the stray field by more than Figure 6 Z-Gradient field strength spectral contours, measured with white noise as indicated in Figure 4. Top: contours of special gradient coil with eddy current screen (db ref 1 [T/A]). Middle: contours of standard gradient coil (db ref 1 [T/A]). Bottom: Difference of magnetic stray fields (db). 2 db. The Y-gradient is shielded by approximately 1 db or more in this region. The stray field at radii smaller than 35 cm is the direct field of the gradient coil, and is therefore hardly influenced by the presence of the eddy current screen on the gradient coil outer diameter and gradient coil flange. Moreover, this strayfield is not relevant for
the development of magnetically induced vibrations in the main magnet outer vacuum chamber. Additional experiments were performed to investigate the magnetic field strength between the gradient coil and the inner bore of the main magnet outer vacuum chamber, which are responsible for the development of magnetically induced vibrations of this part of the outer vacuum chamber. To this end a set of moveable voice coils was put in the space between coil and bore, one coil measuring in radial direction, and one in axial direction with respect to the gradient coil. Because of the limited space available, the axial coil had a sensitivity of 1/1 th of the radial coil. This led to a slightly worse signal to noise ratio for the measurements with the axial coil. It is interesting to note that the inner magnet bore (electrically conducting in this case) exhibits a shielding effect as well as the copper shield. This leads to a relative increase in axial magnetic field, as well as a decrease in radial magnetic field when measured near the surface of the magnet bore. If the magnet bore would be a superconductor the radial field would go to zero, while the axial field would double. The experiment lay-out is shown in Figure 7 p [N/m2] 25 2 15 1 5-5 -1-15 -2 Magnetic pressure on the bore due to eddy of Z-gradient..1.2.3.4.5.6.7.8 Z-position along the bore [m] Figure 8 Calculated magnetic stress on warm bore due to z-gradient for a standard coil. Measurements of the axial magnetic field for the case of Z-gradients are shown below in Figure 9. The different lines correspond to different coils (A-E) which are placed along different positions in the magnet bore, see Figure 7. B(T)/A 1-5 Z-current to magnetic field strength axial coils 1-6 1-7 a b c d e 1-8 1 2 1 3 2 1 phase -1-2 1 2 1 3 Hz Figure 9 axial field generated by Z Figure 7 Magnet outer vacuum chamber (front view, left side view) showing placement of voice coils: A-D fixed axial coils, E moveable axial coil. Calculations were performed for a similar cylindrical 1.5T MR system. This yielded the following pattern of eddy current induced stress along the length of the magnet outer vacuum chamber bore for the case of Z- gradients. These data can be converted to pressure such as shown in Figure 8 with the following equation: p = j B ϕ mz 2Bgz B µ mz (1) Where B gz is the axial stray field and B mz is the static axial field, J φ is the eddy current density in the bore at the point of consideration. This equation has been derived under the assumption that the magnet bore behaves like a superconductor, i.e. that it blocks all incident magnetic field. Performing the calculation gives the following results:
N @ 6A Z-current to magnetic pressure on warm bore tangential coils 1 4 1 3 1 2 a b c d e N/m 2 1 1 1 1 2 1 3 2 15 1 5 phase -5-1 -15-2 1 2 1 3 Hz Figure 1 Calculated magnetic pressure on warm bore due to Z current based on measured magnetic field shown in Figure 9 We see that the DC value based on the measurement with coil A (near the front of the bore) of around 2N/m 2 corresponds well with the peak value of 23N/m 2 calculated near the front of the bore, for the same gradient field strength. Also we clearly see the result of the copper shielding; the magnetic field falls off at frequencies above 5hz or so. At 1khz the field has been attenuated by more than 2dB. The measurements performed with coils 5/E yield more spatial insight, the results are shown below. The measurements were performed by moving the voice coils from the back of the magnet bore to the front in steps of 5cm. Figure 12 Axial magnetic field due to Z gradient Figure 13 Radial magnetic field due to Y gradient Figure 11 Axial magnetic field due to Y gradient Figure 14 Radial magnetic field due to Z gradient Again, we clearly see the frequency-dependent shielding of the copper shroud. Also clearly visible is the aforementioned lower signal-to-noise ratio of the axial coils. When we look at Figure 14 from a different angle we see a position dependency which looks like Figure 8.
3 Summary Figure 15 Same data as in Figure 14, from a different perspective. 2.3 Noise Measurements Many noise measurements were performed but they are not the focus of this paper. A summary of the measurements relevant to the experiments described in this paper is given in Figure 16 db(a) 8 75 7 65 SPL at operator position 125A rms pink noise excitation y-gradient standard (with covers) rigid+shielding+pipe compliant+shielding+pipe compliant+shielding+pipe+covers compliant+shielding+end plates+covers The significance of the magnetic transmission path of gradient coil to main magnet vibrations has been shown to play a significant part in the noise production of an MRI system. Without addressing this transmission path noise reduction of more than 1dB(A) is very difficult. A novel and patented method [2] of shielding the main magnet outer vacuum chamber from stray magnetic field originating from the gradient coil has been demonstrated. The effectiveness of the shielding is about 2dB when looking at main magnet bore excitation, and about 1dB when looking at the front and back sections. The noise reduction that was measured on the system was in line with this. References [1] Price, De Wilde, Papadaki, Curran, Kitney, Investigation of Acoustic Noise on 15 MRI Scanners from.2t to 3T, Journal of magnetic resonance imaging 13: 288-293 (21). [2] Patent WO-/25146, MRI apparatus with a mechanically integrated eddy current shield in the gradient system, Priority date 28 October 1998, Filing date 4 October 1999, Publication date 4 May 2. 6 55 5 1 2 Figure 16 Summary of noise measurements at a position in front of the scanner, not directly in front of the bore. These measurements show that the magnetic shielding in itself does not yield noise reduction, because the structural transmission path dominates. Introducing a compliant gradient coil suspension and main magnet covers lowers the noise level to about 1dB(A) below that of a standard scanner. The fact that the covers help significantly indicates that main magnet vibrations are contributing significantly. Adding copper endplates to the gradient coil (effectively closing it off acoustically as well as magnetically) significantly lowers the noise level to a level that lies 2dB(A) below that of a normal scanner. This indicates that without the endplates, the magnetic transmission path is still significant.