ArticlePDF Available

Magnetocardiography under Clinical Conditions

De Gruyter
Biomedical Engineering / Biomedizinische Technik
Authors:
Magnetocardiography under Clinical Conditions
Erné, S.N.
1
, Kammrath, H.G.
1
, Ziolkowski, M.
2
, Tenner, U.
1
, Müller, H.-P.
1
1
Division for Technical Bioelectromagnetism, Central Institute for Biomedical Engineering,
Ulm University, D – 89069 Ulm, Germany
2
Szczecin
Technical University, Al. Piastow 19, PL - 70310, Szczecin, Poland
Introduction
During the last years several algorithms and analysis
procedures were developed for the evaluation of
biomagnetic and bioelectric data. To obtain clinical
validation, an Magnetocardiography (MCG) system
especially designed for clinical studies on large patient
number has been installed in Ulm (Figs. 1-3) [1].
Fig. 1: Global view of the facility. The shielding
performance is described in [2].
.
Fig. 2: MCG measurement device.
Fig. 3: Operator Console
Standardized application software has been developed
that simultaneously combines and merges established
time series algorithms as well as spatio-temporal
analysis algorithms from high resolution
electrocardiogram (HR-ECG) with their extensions to
MCG data.
Method
The MCG system installed at the Ulm University is
specifically designed for clinical application and routine
use and this implies that a large number of patients are to
be investigated. To reach this goal, the system design
meets the requirements of reliability and high field
sensitivity. The MCG sensor system is operating inside
the magnetic shielded room described in [2]. The sensor
system consists of a planar dewar, containing a complex
architectural structure with sensors distributed over three
levels. The first level, i.e. the primary measurement
plane holds 55 SQUID sensors. The sensing elements
are integrated magnetometers with a square shape of
12.7 mm in diagonal. The sensors are uniformly
distributed over the inner surface of the dewar,
according to a hexagonal geometry, covering a circular
surface of about 230 mm in diameter. The measurement
plane is 18 mm from the outer dewar bottom. Nineteen
additional SQUIDs are mounted on the second level and
are used as reference channels 90 mm from the
measurement plane. On the third level there is a
magnetometric triplet located 70 mm from the reference
plane to control the active shielding system. Operating in
the shielded environment the system shows on all
channels a white noise level better than 10 fT/Hz at 10
Hz. A software-gradiometer set-up is easily performed
Biomedizinische Technik, 1999, Vol. 44, No. 2, pp. 152-155
by subtraction of the background field sensed by
selected reference channels from the signal of each
primary channel.
The patient handling set-up is made of non magnetic
material and is an emulation of the standard set-up for
clinical cardiac electrophysiology (fig. 4).
For cardiac measurements a map of 55 magnetic
channels, three orthogonal electrocardiographic (ECG)
leads and a breathing channel are used.
For a raw patient positioning a set of laser markers is
used. These markers are set up to point on positioning
coils on the ends of a cross which is fixed at the sternum.
The dewar is tilt at an angle of 10° to get aligned to the
thorax surface. The measurement system is positioned at
a distance of about 10 mm from the thorax. This
procedure allows for a position of the central sensor
approximately 120 mm caudal of manubrium sterni and
50 mm left of the sternum, so that mainly the field
normal to the thorax surface is measured. The patient
positioning unit (PPU) determines the patient position
relative to the sensor array using three coils on the cross
that is fixed along the sternum of the patient. The PPU
measurement takes only few seconds before the five
minutes MCG measurement starts. As example for the
55-channel MCG, the normal heart beat of a volunteer is
illustrated in fig. 4.
Fig. 4: Example of an averaged heart beat (100 beats).
The signals were recorded simultaneously at 55
positions.
Digital filtering at a cut-off frequency of 250 Hz is
applied by digitizing the MCG signals at a sampling rate
of 1 kHz.
The recorded raw data are stored on compact discs and
afterwards are automatically preprocessed by the
analysis software, described below.
The MRT data are needed for an accurate representation
of the individual geometry of the thorax and the heart
surfaces for volume conductor modeling. The MRT
images are recorded in a superconducting magnet device
(Magnetom Vision 1.5 T Siemens, Erlangen) using a
body array coil. Electrocardiographic leads used for
cardiac gating. With a normal T
1
-spin-echo-sequence,
implemented as a standard technique of the Siemens
software pulse-sequences, transverzal images are
obtained from apex to base in order to cover the entire
heart and the thorax. For a useful volume conductor
modeling 64 slices in transverzal direction with
extension 512 x 512 pixels should be detected. Then, the
voxel size would be about 1 x 1 mm
2
in the transverse
slices with a slice thickness of about 5 mm. Thus, a field
of view of 51.2 x 51.2 cm
2
in the transverzal slices and
32.0 cm perpendicular is recorded. With an echo time T
E
of about 15 ms and a repetition time T
R
> 800 ms,
according to heart beat gating a total measurement time
of about 15 min can be reached.
Fig. 5: Channel display for ECG signals. The 3
orthogonal leads are displayed.
At the beginning the Biomagnetic Analysis Console
(BAC) software was used for MCG signal analysis [3].
Now, the standard analysis tool is the Open Magnetic
and Electric Graphic Analysis (OMEGA) software [4].
The OMEGA software contains established time series
analysis algorithms with their extensions to MCG and
ECG (Figs. 5,6) data as well as tools for the localization
of cardiac electromagnetic sources.
Biomedizinische Technik, 1999, Vol. 44, No. 2, pp. 152-155
Fig. 6: Multichannel display for 55 MCG signals.
The OMEGA software is using different analysis
algorithms under exactly defined mathematical
conditions on the same data set to pursue synergetic
interactions. Therefore, medical information for clinical
applications and diagnostics can be obtained.
The OMEGA software was developed object oriented to
run on UNIX workstations as well as on PCs. To obtain
flexibility and expandability the program is divided into
different parts:
reader tools for MCG or high resolution ECG data
sets
preprocessor tool
analysis tools
reader tools for CT or MRT images
2D/3D visualization tools
graphic user interface
As analysis methods are included frequency analysis,
binomial filtering, Simson analysis, heart beat
variability, principle component analysis, smoothness
test and further analysis algorithms (Figs. 7-10) as well
as the visualization of the field distribution (Fig. 7).
Fig. 7: Averaged data display. The MCG and ECG
signals are displayed as overlay plots. On the right side
a contour plot of the magnetic field for a defined time
point (see the line in the overlay plot) is displayed.
Fig. 8: Example for a time series analysis tool:
Frequency analysis.
Spatio-temporal analysis and current source localization
methods are implemented as well. These methods are
using localized (dipole) sources or distributed sources.
Tools for generating 3D models obtained from MRT are
included, 2D/3D visualization of the results of the
electromagnetic source reconstruction in the
morphologic environment is also possible (Fig. 11).
Fig. 9: Example for a time series analysis tool: Binomial
filtering analysis.
Fig. 10: Example for a time series analysis tool: Simson
analysis.
Visualization in medicine is usually linked up to CT-or
MRT-images. The information received using these
techniques can be displayed as a set of 2D images or
combined 3D blocks with volumetric rendering methods.
In numerical modeling of biomagnetic fields in the
human body, the basic goal is not only to show the
anatomical structure but also to present calculated
electromagnetic fields and localized sources in the
morphological environment.
OMEGA provides the possibility to present calculated
fields, i.e. the electromagnetic information, in
combination to the anatomy. The working scene for the
presentation of the electromagnetic fields and the
anatomical objects (boundary element method (BEM)
meshes, MRT slices, CT slices) is a rotating 3D-cube
Biomedizinische Technik, 1999, Vol. 44, No. 2, pp. 152-155
controlled by so called virtual trackball, which enables
the rotation in any direction. The scalar fields can be
displayed in a form of isocontours or equifilled regions
and the vector fields can be shown as a set of solid
cones.
Common dipole reconstruction methods work with non-
linear least-squares parameter estimations for solving the
inverse problem. This method needs costly computations
if instead of an analytical volume conductor (sphere,
halfspace, ellipsoid, etc.) realistic geometry models are
involved.
Fig. 11: Visualization of the localization results on a
BEM modeled torso.
The BEM torso consists of some
500 triangles.
The OMEGA software contains a linear iterative
algorithm for dipole localization. This algorithm
provides the possibility of computing dipole localization
in analytically described surfaces as well as in realistic
triangulated surfaces in a time which is in a useful range
for routine applications of biomagnetism. The method is
described in [5].
The goal in solving the biomagnetic inverse problems,
is to calculate the source current density distribution
producing the magnetic signals measured outside the
body. The method that is implemented in OMEGA to
realize this task is based on the minimum-norm
approach [6]. Weighting matrices to eliminate depth bias
of localized sources and to correct the lead field matrix
with the field deviations [7] are used. The Tikhonov-like
regularization [8] is used in the process of solution for
balancing noise influence. The source space where the
current distribution is reconstructed can be defined as a
regular 3D grid of current dipoles locations or can be
described as a 2D surface arising from physiological
constrains.
Results
Within a training study more than 100 MCG
measurements have been performed. The volunteers
have been selected having no reported cardiac diseases.
During the course of the training study the measurement
protocol was iteratively improved. An optimized
recording strategy and an optimized measurement
position for the patient has been defined and can be set
up within a few minutes, so that an MCG measurement
can be performed within 10 minutes.
Discussion
The OMEGA software has been developed to provide an
easy, reproducible and fast analysis and visualization
possibility for electric and magnetic data sets in
combination with the image data sets of the
corresponding anatomy.
A special feature of OMEGA is the open structure
prepared for implementation of further analysis
algorithms and new techniques of visualization.
When clinical measurements are considered there is a
need to observe a large number of patients and thus short
measurement time, fast data analysis and large storage
capability. A validation of the results of magnetocardio-
graphy at a clinical level, allowing a clinical routine use
of MCG in the contest of clinical work in cardiology is
on the way. A clinical system has to perform a standard
MCG measurement within a global time of about 15
minutes per patient, leading to a possible throughput of
30 patients per day. The results of the study on more
than 100 subjects to iteratively improve the
measurement protocol and the system logistic have
shown that these performances are realistic for the Argos
55 system. Of course, for a clinical validation the
analysis routines have to be tested on a large data base.
References
[1] Erné, S.N., Müller H.-P. Magnetocardiographic
technology: state of the art, Proceedings of 8
th
International
Congress on Holter and Noninvasive Electrocardiology,
ISHNE98, Ulm, Germany, 1998.
[2] Pasquarelli, A., Kammrath, H., Tenner, U. and Erné, S.N.
The new Ulm magnetic shielded room, Proceedings of
BIOMAG98, Sendai, Japan, 1998.
[3] Martin, K., Erné, S.N., Law, C., Mallick, J., Tatar, B.,
Multiple modality biomagnetic analysis system, 10
th
International Conference on Biomagnetism, BIOMAG ‘96,
Santa Fe, USA, 1996.
[4] Tenner, U., Kammrath, H., Erné, S.N., OMEGA Open
Magnetic and Electric Graphic Analysis, 8
th
International
Congress on Holter and Noninvasive Electrocardiology,
ISHNE98, Ulm, Germany, 1998.
[5] Del Gratta, C., Erné, S.N., Edrich, J. A Linear iterative
algorithm for dipole localization. In Baumgartner, C., et al.,
Biomagnetism: Fundamental research and Clinical
Applications, Elsevier Science, 1995.
[6] Nenonen, J.T., Hämäläinen, M.S., Ilmoniemi, R.S.,
Minimum-norm estimation in a boundary-element torso model,
Medical & Biol. Engineering & Computing, 43-48, 1994.
[7] Brauer, H., Kosch, O., Tenner, U., Arlt, A., A modified
linear estimation approach for solving biomagnetic inverse
problems. IEEE Transactions on Magnetics, 32, 1298-1301,
1996.
[8] Tikhonov, A.N., Arsenin, V.Y., Solutions of Ill-possed
Problems, New York, Wiley, 1977.
Biomedizinische Technik, 1999, Vol. 44, No. 2, pp. 152-155
... In the work presented here, a new software platform -Open Magnetic and Electric Graphic Analysis (OMEGA) [1] -has been developed that includes the analysis of fMRI image series, tools for volume modeling of MRI data and the localization of functional human brain areas by solving the inverse problem for MEG data sets. The fast tools and algorithms enable a complete and user-independent analysis to compare and integrate the localization results of brain mapping with co-registered different modalities: ...
Article
Full-text available
Introduction During the last years, interest has grown in novel methods for mapping human brain function. The relevant techniques are functional magnetic resonance imaging (fMRI), magnetoencephalography (MEG). The comparison of those different techniques can be performed by multimodal mapping to test the validity of the different modalities in one analysis environment. Moreover, multimodal mapping increases the reliability in the results of a single modality. In the work presented here, a new software platform -- Open Magnetic and Electric Graphic Analysis (OMEGA) [1] -- has been developed that includes the analysis of fMRI image series, tools for volume modeling of MRI data and the localization of functional human brain areas by solving the inverse problem for MEG data sets. The fast tools and algorithms enable a complete and user-independent analysis to compare and integrate the localization results of brain mapping with co-registered different modalities: . fMRI analysis and vis
Article
High resolution electrocardiography (HRECG) recordings have already shown an increased beat-to-beat microvariability of the QRS duration of the terminal QRS in patients with a history of ventricular tachycardia (VT). The purpose of this study is to detect QRS-duration microvariability with magnetocardiographic (MCG) recordings in normals, patients with coronary heart disease (CHD), patients with a history of myocardial infarction (MI), and VT patients. QRS microvariability is calculated as the variance of time-shifts of single beats respectively to the average of all beats. The average over all channels of the MCG is performed. QRS microvariability was evaluated from 55-channel MCG in 15 normal persons, in 12 patients with CHD, in 13 patients with MI, and in 10 patients with VT. We found a significantly higher microvariability in patients with MI compared to normals. The highest microvariability was found in VT patients.
Article
Parameterization of the ST-segment is used as a tool for risk stratification for patients to suffer from ventricular tachycardia. This parameterization is performed in terms of Principal Component Analysis (PCA) applied on multichannel magnetocardiographic (MCG) recordings. 55-channel MCG was recorded from 14 normal persons, 10 patients with CHD, 14 patients with MI, and six patients with VT. We found a significantly (p < 0.05) lower PCA-score in patients with MI compared to normals. The lowest PCA-score was found in VT patients. Significant differences can be found between VT patients and normals and also between VT patients and CHD patients.
Article
Performing signal averaging in an efficient and correct way is indispensable since it is a prerequisite for a broad variety of magnetocardiographic (MCG) analysis methods. One of the most common procedures for performing the signal averaging to increase the signal-to-noise ratio (SNR) in magnetocardiography, as well as in electrocardiography (ECG), is done by means of spatial or temporal techniques. In this paper, an improvement of the temporal averaging method is presented. In order to obtain an accurate signal detection, temporal alignment methods and objective classification criteria are developed. The processing technique based on hierarchical clustering is introduced to take into account the non-stationarity of the noise and, to some extent, the biological variability of the signals reaching the optimum SNR. The method implemented is especially designed to run fast and does not require any interaction from the operator. The averaging procedure described in this work is applied to the averaging of MCG data as an example, but with its intrinsic properties it can also be applied to the averaging of ECG recording, averaging of body-surface-potential mapping (BSPM) and averaging of magnetoencephalographic (MEG) or electroencephalographic (EEG) signals.
Chapter
As an emerging new modality, biomagnetic analysis presents exciting opportunities while raising some difficult issues. During the past four years researchers at General Electric, The University of Ulm and The University of Chieti have created a state of the art biomagnetic analysis system, and have addressed many of these difficulties. One of the critical aspects of a biomagnetic analysis system is how it handles different types of data. Many of the clinical requirements for a successful system require fast, efficient and flexible data management. This paper describes what those requirements are and how we addressed them. We start by providing a high level overview of our system. We then list some of the pertinent clinical requirements, present our solutions and describe how well they have worked. Finally we discuss future directions.
Chapter
The non-invasive detection of high resolution magnetocardiographic signals generated by the electrical activation of the heart requires the development of high-resolution magnetic field sensors (superconducting quantum interference devices (SQUlDs)). The ability of multichannel recording and adequate computational power allows an application of this technique suitable for clinical use. The magnetic shielding of the detection chamber is a necessity for a high signal to noise ratio. The basic technical performances, an overview of large MCG systems and as an example for a clinical MCG system a description of the performance of Argos 55 — the new MCG system in Ulm — is presented.
Article
A computer simulation study of the biomagnetic inverse problem was performed. A realistically shaped, multi-compartment boundary element model of the human body was used to localize dipolar current sources within the heart. The minimum-norm solution of the often used Linear Estimation Approach (LEA) was modified to improve localization accuracy. By multiplying the lead field matrix with two new weighting matrices and implementing a procedure for shifting the reconstructed current dipoles the authors could avoid the well-known blurring effect which can be observed if the common LEA is used. The new modified linear estimation approach (MLEA) was applied to localization of different dipole configurations. It can be used to reconstruct volume sources as well
Article
The paper deals with the bioelectric and biomagnetic inverse problems. The authors present a method to estimate primary-current distributions in a homogeneous, realistically shaped boundary-element torso model. The reconstruction surface is triangulated to keep the procedure computationally feasible. The minimum-norm estimate is computed on the basis of separate electric and magnetic signals, as well as from combined data. The method can be used both for heart and brain studies. Simulation results for current-dipole sources in a homogeneous realistic torso are discussed.
The new Ulm magnetic shielded room
  • A Pasquarelli
  • H Kammrath
  • U Tenner
  • S N Erné
Pasquarelli, A., Kammrath, H., Tenner, U. and Erné, S.N. The new Ulm magnetic shielded room, Proceedings of BIOMAG98, Sendai, Japan, 1998.
A Linear iterative algorithm for dipole localization
  • C Del Gratta
  • S N Erné
  • J Edrich
Del Gratta, C., Erné, S.N., Edrich, J. A Linear iterative algorithm for dipole localization. In Baumgartner, C., et al., Biomagnetism: Fundamental research and Clinical Applications, Elsevier Science, 1995.
OMEGA Open Magnetic and Electric Graphic Analysis
  • U Tenner
  • H Kammrath
  • S N Erné
Tenner, U., Kammrath, H., Erné, S.N., OMEGA Open Magnetic and Electric Graphic Analysis, 8 th International Congress on Holter and Noninvasive Electrocardiology, ISHNE98, Ulm, Germany, 1998.