IEEE TRANSACTIONS ON APPLIED SUPERCONDUCIIVIR, VOL. 5, NO. 2, JUNE 1995
Active Noise Compensation for Multichannel
Magnetocardiography in an Unshielded Environment
W.A.M. Aarnink, P.J. van den Bosch, T.-M. Roelofs, M. Verbiesen, H.J. Holland, H.J.M. ter Brake and H. Rogalla,
University of Twente, Applied Physics D e p a r t m e n t , Low Temperature Division, P.O. Box 217,7500 AE Enschede, The Netherlands
Abstracf-A multichannel high-T,-SQUID-based heart
scanner for unshielded
development. Outside a magnetically shielded room,
sensitive SQUID measurements are possible using
gradiometers. However, it is difficult to realize large-
baseline gradiometers in high-T, materials. Therefore, we
developed two active noise compensation techniques. In
the Total Field Compensation technique, a Helmholtz
type coil set is placed around the sensors. One
magnetometer is used as a zero detector controlling the
compensation current through the coil set. For Individual
Flux Compensation, the reference signal is sent to the
separate SQUIDS (or their flux transformer circuits) to
compensate the local environmental noise fluxes. The
latter technique was tested on low-T, rf-SQUID
magnetometers, each sensor set to a field resolution
typical for high-T, SQUID magnetometers, i.e.
0.1 pTddHz. We were able to suppress the
environmental disturbances to such an extent that
magnetocardiograms could be recorded in an ordinary
environment. In the paper the two suppression techniques
are described and experimental results are presented.
environments is under
Up to now, superconducting quantum interference devices
(SQUIDS) are the most sensitive magnetometers. Only with
these devices biomagnetic signals (human heart, brain) can be
detected. Low-T, SQUIDs have been widely applied for
encephalography (MEG) [ 1 - 31.
Presently we a i m at the development and realization of a
high-T, SQUID based prototype heart scanner. The system
should detect the biomagnetic signal of the (fetal) human
heart in a magnetically unshielded environment (e.g. in a
clinic or hospital). The system consists of high-Tc SQUID
magnetometers, a cooling subsystem and an environmental
noise suppression subsystem.
The biomagnetic signal level of the fetal human heart is
1 - 10 pT, and of the adult human heart 10 - 100 pTW [4,5].
If we want to perform magnetocardiography (MCG) in a
frequency range of 0.1 - 40 Hz, we need a magnetometer
resolution of about 0.1 pT-/dHz.
adult and fetal, is very small compared to the magnetic noise
The heart signal, both
Manuscript received October 18, 1994.
Supported by the Dutch National Research Program on High-Tc
Superconductors and Philips GmbH Research Laboratories, Hamburg, FRG.
level in a typical laboratory or clinical environment (low
frequency fluctuations in the order of a few nT, about 10 nTpp
at 50 Hz). Without precautions, the signal-to-noise ratio is far
too low for biomagnetic measurements in an unshielded
11. PROBLEM DESCRIPTION
For our purposes, a magnetically shielded room (MSR) is
not usefd. It is expensive and hardly transportable.
Therefore, other ways to suppress environmental disturbances
have to be developed.
One possibility to overcome problems with external
magnetic fields would be the measurement of the magnetic
field gradient instead of the magnetic field itself. Using
superconducting wire one can build gradiometers measuring
the gradient of the magnetic field and therefore, suppressing a
uniform noise field fiom the environment. Employing a
second order gradiometer, one is able to perform biomagnetic
measurements in a magnetically unshielded environment .
High-Tc SQUIDs have the advantage of a relatively high
operating temperature. As a cooling agent, liquid nitrogen
instead of liquid helium can be used. Liquid nitrogen has the
advantages that it is cheap and relatively easy to handle.
However, large-baseline gradiometers in high-Tc materials
cannot yet be realized in a simple manner. Therefore, active
noise compensation techniques have to be incorporated in our
high-Tc SQUID based heart scanner.
111. ACTIVE NOISE COMPENSATION TECHNIQUES
Active noise compensation techniques were applied before
[7 - 111. We tested two techniques by means of flux gate
magnetometers (FGMs) and low-T, rf-SQUIDS. The a i m was
to investigate whether these techniques could be applied for
high-Tc SQUID magnetometers. In this paper the two
techniques are described and noise measurements as well as a
magnetocardiogram are presented and discussed.
A. Requirements for noise compensation techniques
Above we already saw that the environmental disturbances
are much stronger t h a n the magnetic signal of the human
heart. Environmental disturbances have to be suppressed
40 - 60 dB for magnetocardiography (MCG) and 60 - 80 dB
for fetal magnetocardiography (FMCG).
1051-8223/95$04.00 0 1995 IEEE
C. Individual Flux Compensation
Fig. 1. Total field compensation (TFC).
B. Total Field Compensation
Total field compensation (TFC) can be applied for the
improvement of a p-metal magnetically shielded room .
This technique may also be used to compensate disturbing
environmental magnetic fields, see Fig. 1. Inside a large
Helmholtz coil set an array of SQUID magnetometers is
placed. One sensor is used as a zero detector controlling the
compensation current through the coils. We tested this
compensation technique using FGMs. Suppression factors of
20 - 30 dE3 were observed. One of the disadvantages is the
difficulty to produce a uniform compensation field near the
sensors. For a large sensor array this leads to large
compensation coils, decreasing the transportability of the
system. Therefore, TFC is less suited for our heart scanner.
SQUID electronics .
SQUID array F'"
the magnetic-field-to-voltage transfer function we find:
AVIAB = 3.9.107 V/T. Above we used theoretical values to
calculate AVIAB, in practice a relative error of about +20%
may well occur in AVIAB, due to errors in AVIAOSQ and M.
Fig. 2. Individual flux compensation (IFC).
Instead of using a large Helmholtz set, one also can use
separate small coils near the magnetometers to locally
suppress environmental disturbances, see Fig. 2. The output
of one of the channels is used as a noise reference and sent to
the separate SQUIDs (or their flux transformer circuits) to
compensate the local environmental noise fluxes. Therefore,
this technique is called individual flux compensation (IFC).
Compensation coils and magnetometers can be integrated e.g.
in a small tube, enabling the development of a modular
system. In such a system, channels can easily be added and
serviced, if necessary. Also the dimensions of the system can
be smaller, favouring a transportable set-up. This technique
was tested with low-T, rf-SQUIDS, as is discussed in the
Iv. APPLICATION OF INDrvIDUAL FLUX COMPENSATION
A. Experimental set-up
In the future, we want to apply IFC on high-T, SQUIDS.
We developed our magnetometers, which are based on low-T,
SQUIDs, in such a way that their intrinsic noise levels are
comparable to high-T, SQUID magnetometers. Our results
then are indicative of what can be achieved with the
combination of high-T, SQUIDs and IFC.
The IFC technique was tested with two low-T, rf-SQUIDS
[ 121 each connected to a magnetometer-type pick-up coil, see
Fig. 3. The distance or baseline b between the two pick-up
coils was about 10 cm. The transfer of each flux transformer
was designed in such a way that the field resolution of the two
channels equals the typical value of high-Tc dc-SQUID
magnetometers of 0.1 pTddHz at 77 K [13, 141. However,
one should note that much lower values can be reached when
applying low-T, rf-SQUIDS (e.g. see [ 151). For the magnetic-
field-to-voltage transfer function
magnetometers we may write:
AV AV ADsQ AQp
AB A@,, ABp AB '
where OsQ is the flux in the SQUID and Op the flux in the
pick-up coil. For the SQUIDs we used, the flux-to-voltage
transfer function AV/A@Q equals 20 mVI+, , with $0 the
flux quantum. The flux transformer consisted of 3 pick-up
windings with a radius of 7 mm, inductively coupled to the
SQUID via a mutual inductance M. In our case M equals
20nH, whereas the total flux transformer inductance Lf
equals 2.4 pH. Therefore: AOs~ACDp = MILf = 8.3.10-3. The
transfer function AOplAB equals Ap, the total area of the 3
pick-up windings (in our case, Ap = 4.62 cm'). Therefore, for
AVIAB of the
- - -- -'
Fig. 3. Experimental set-up for testing IFC on low-T, rf-
In IFC, the signal of the reference channel is inductively
fed into the flux transformer circuit of the measuring channel
via a compensation resistance &, see Fig. 3. The mutual
inductance & that was used for this purpose consisted of a
horse shoe shaped toroid (radius 4 mm) with a primary and a
secondary coil of 100 and 20 windings, respectively. Each
winding had a radius of 1 mm. The inductance LCs of the
primary coil equaled 2.106H, the inductance L c ~
secondary coil was 8-10-*
H and the mutual inductance It4~
was about 4 . 1 0 - ' H (with the coupling constant k near 1 ) . For
the compensation current IC used for flux compensation, we
may write, neglecting the impedance Lc.s of the primary coil:
where BENV is the environmental magnetic noise field. For
optimal compensation (1st order gradiometer) the flux OC
should equal the flux Op = ApBW in the pick-up coil of the
signal line and we find:
yielding a value of 33 WZ for &. Again we used theoretical
values to calculate &. Above we saw that in AV/AB alone a
relative error of f20% may occur. Therefore, relative errors
in & of about f30% may be expected, also due to an error in
B. Noise considerations
For the SQUIDS we used, the intrinsic SQUID flux noise
level (S@,SQ)" equaled 2-104 t$ddHz . For the equivalent
flux noise (Sm,p)'R in the pick-up coil we then may write:
and we find (Sm,p)'n = 2.4-10-2
t$~/dHz. This corresponds to a
field noise level (SB)In = (S,,p)lR/Ap of 0 . 1 pT~&h-Iz for
each magnetometer channel.
In IFC, the intrinsic noise of the reference c h q e l is fed
into the measuring channels. For the corresponding flux noise
level (S@,c)'n of the compensation flux coupled into the flux
transformer we find
AV Mc (&)'I2
This compensation flux noise adds to the intrinsic noise of
the measuring channel. Since these are not correlated, we
find for the theoretical flux noise level (So,IFC)'R in the
(Sm,J2 = ( % , P
and in our case this equals 3.4~10-~
a field noise level (SB,pc)ln of 0. I5 pT&dHz.
+ %cy2 7
t$&kz, corresponding to
C. Optimization of compensation settings
For optimization and testing purposes, we put our set-up in
a large Helmholtz coil set-up (coil diameter 1.8 m). We
subjected our SQUID magnetometer system to a uniform
harmonic magnetic field with a fiequency of about 5 Hz. The
compensation resistance & was tuned so that the amplitude
of the resulting field at this fiequency after compensation was
minimal. With & = 43 WZ we found suppression factors for
the 5 Hz magnetic field of about 55 dB. After tuning, the
Helmholtz coil set was disconnected. Noise measurements
have been performed with and without IFC, see Fig. 4 . In a
range of 0.5 - 20 Hz, suppression factors for the
environmental disturbing field of about 40 dB have been
found. At higher fkequencies suppression factors of 20 - 30
dB have been measured. We see that above 3 Hz the noise
Fig. 4. Noise measurements with (solid line) and without
(dashed lime) IFC.
level is around OSpT~ldHz, a factor of 3 above the
expected intrinsic noise level (SB,Fc)’n = 0.15 pT-/dHz
the system. We speculate that gradients in the noise field
combined with the fact that the two pick-up coils are not
exactly coplanar, are the cause for this unexpected increase.
Magnetocardiography in an unshielded environment
From Fig. 4, we see that by applying IFC we can reach noise
levels of about 0.5 pTMS/dHz. This noise level is low enough
to measure the magnetic heart signal of an adult. In Fig. 5, a
magnetic heart signal as we recorded it, is shown. An analog
2nd order low pass filter with a cut-off ftequency of 10 Hz
was applied, followed by a digital filter removing the
remaining peaks at 50 and 100 Hz.
We successfully applied IFC in combination with low-T,
rf-SQUIDS for MCG in an unshielded environment. However,
the noise level of our sensor is about a factor of 3 above the
expected value. We expect to improve the field sensitivity by
applying advanced gradiometric configurations (implemented
in sohare) and by arranging the magnetometers in a
perfectly coplanar way.
An attractive application of a heart scanner would be fetal
magnetocardiography (FMCG). Application of electrocardio-
graphy is limited for the study of the fetal heart signal .
The magnetic signal level of the fetal heart is about a factor of
10 below that of an adult heart. We showed that with a
resolution of 0.5 pT-/dHz
FMCG we would need a system with a resolution of about 50
ff~sIdHz to obtain a similar signal-to-noise ratio as in our
present MCG. Therefore, also the improvement of the field
sensitivity of high-Tc SQUID magnetometers needs attention,
if one wants to apply these sensors for FMCG.
we can perform MCG. For
0.00 I‘ ‘
Fig. 5. Typical magnetocardiogram, taken in a magnetically
unshielded environment using individual flux compensation.
For large sensor arrays in which IFC is applied, it will be of
great importance that the sensors are coplanar. We did not yet
investigate this in detail. In the hture, the effect of deviations
in this respect will be investigated on a multichannel system.
Our set-up effectively resembles a first order gradiometer.
Higher order gradients of the disturbing environmental field
are not removed. Therefore, nearby noise sources, producing
higher order gradients, will be less effectively suppressed.
This is especially the case when large sensor arrays with large
baselines are used. Future studies also should concentrate on
We used a large Helmholtz coil set to tune our IFC. In a
follow-up system correlation techniques for tuning the
different channels will be applied.
S.J. Williamson, G.-L. Romani, L. Kaufman, and I. Modena, Ed?..,
Biomagnetism an Interdisciplinary Approach, New York: Plenum
 S.J. Williamson, M. Hoke, G. Stroink, and M. Kotani, Ed?..,
Advances in Biomagnetism, New York: Plenum Press, 1989.
 M. Hoke, S.N. Erne, Y.C. Okada, and G.L. Romani, Eds.,
Biomagnetism Clinical Aspects, Amsterdam: Elsevier Science
Publishers B.V., 1992.
 T. Katile and P. Karp, “Magnetocardiography: morphology and
multipole presentations”, in [I], pp. 237-263.
 V. Kariniemi and T. Katile, “ Perinatal magnetocardiography”, in
[I], pp. 299-310.
magnetocardiography 10 years experience at the Catholic
university”, Int. J. ofcardiac Imaging, vol. 7, pp. 151-167, 1991.
 H.J.M. ter Brake, R. Huonker, and H. Rogalla, ‘New results in
active noise compensation for magnetically shielded rooms”,
Meus. Sci. Technol., vol. 4, pp. 1370-1375, 1993.
 K. Fujioka, K. Matsumoto, Y. Yamagishi, T. Noda, and Y.
Kuraoka, “Active shield with
Superconducting Devices and
Applications, H. Koch and H. Lubbig, Ed?.., Berlin: Springer-
Verlag, 1992, pp. 533-536.
 H. Seifert, A. Wirth, and S. Schneider, “Active magnetic shielding
support for biomagnetic instruments”, in Superconducting Devices
and Their Applications, H. Koch and H. Lubbig, Ed?.., Berlin:
Springer-Verlag, 1992, pp. 537-541..
[IO] K. Yokosawa and S. Kuriki, “Design and fabrication of direct-
feedback gradiometer”, to be published in the Proc. of the 9th Int.
Cod. on Biomagnetism, Vienna, 1993.
[Ill R.H. Koch, J.R. Rozen, J.Z. Sun, and W.J. Gallagher, “Three
SQUID gradiometer”, Appl. Phys. Lett., vol. 63, pp. 403-405,
[I21 Biomagnetic Technologes Inc., PO Box 210079, San Diego, CA
92121, USA. System 330 x.
[I31 J.W.M. Hilgenkamp, G.C.S. Brons, S. Hoogeveen, J. Flokstra, and
H. Rogalla, “Monolithic flux transformer coupled high T, dc
SQUID magnetometers”, to be published in IEEE Trans. on Appl.
[I41 D. Koelle, A.M. Micklich, F. Ludwig, E. Dantsker, D.T. Nemeth,
and J. Clarke, “dc SQUID magnetometers from single layers of
YBa2Cu307.,”, Appl. Phys. Lett., vol. 63, pp. 2271-2273, 1993.
[I51 H.J.M. ter Brake,.P.J. van den Bosch, and H.J. Holland, “Magnetic
noise of small Stirling coolers”, in Advances in Cryogenic
Engineering, vol. 39B,, P. Kittel, Ed., New York: Plenum Press,
1994, pp. 1287-1295.
Fenici, G. Melillo, and M. Masselli, “Clinical
SQUID for biomagnetic
[I61 Z. Dunajski and M. Peters, “Volume current effects on the fetal
magnetocardiogram, in , pp. 565-569.