A PHANTOM STUDY ON THE EFFECT OF A THIN INHOMOGENEITY LAYER ON BODY SURFACE POTENTIAL MAPS

M. Lindholm^1,2, J. Nenonen^1,2, M. Karvonen^1,2, M. Liehr³, J. Schreiber³, J. Haueisen³,
and T. Katila¹
¹Laboratory of Biomedical Engineering, Helsinki University of Technology, P.O. Box 2200, 02015 HUT, Espoo, Finland
²BioMag Laboratory, Helsinki University Central Hospital, P.O.Box 509, 00290 HUS, Helsinki, Finland
³Biomagnetic Center Jena, Friedrich Schiller University, D-07743 Jena, Germany

Abstract: We performed phantom recordings to test how much an artificial PVC sternum and thin inhomogeneity sheet mimicking the skeletal muscle layer affect body surface potential maps generated by a current dipole. Inverse calculations with an equivalent current dipole in a boundary-element torso showed an excellent accuracy for the homogeneous phantom. When the inhomogeneities were inserted the potential patterns did not exhibit significant change, but the calculated dipole depth increased significantly for dipoles lying close to the inhomogeneity.

INTRODUCTION

Human body is often modeled as being homogeneous and isotropic in its electrical properties. However, in some tissue types the conductivity is strongly direction-dependent. Myocardium, brain and ordinary muscle tissue are examples of such an anisotropy. In the myocardium, as well as in ordinary muscles, the anisotropy is due to the cell structure of the muscles, which favors the electric conduction into the direction of the cells and hinders the conduction in the perpendicular directions.

In patient studies at Helsinki University Central Hospital (HUCH), we discovered that magnetocardiographic (MCG) measurements were able to locate dipolar non-magnetic pacing catheters in the right ventricle with the accuracy of 7 mm, while simultaneous body surface potential mapping (BSPM) was placing the dipole systematically almost 20 mm too deep [1]. The hypothesis is that the horizontally layered anisotropic skeletal muscle layer distorts the potential of the dipole while it has no effect on the magnetic field on the thorax surface.

In the present study we performed phantom experiments to study how much an artificial sternum and a layer with differing conductivity affect the BSPM patterns and dipole localizations. Even though the layer, mimicking skeletal muscle, is not anisotropic, it should qualitatively reflect the effects of the inhomogeneous and anisotropic conductivity between the sources and the sensors.

METHODS

Measurements

Phantom measurements were performed both with the 99-channel cardiomagnetometer and the 123-channel BSPM system [2] in the BioMag Laboratory at HUCH. In this paper we report only the BSPM results.

Thorax phantom and dipole sources

A fiberglass phantom with the shape of a truncated adult male torso was employed (for details, see http://www.biomag.uni-jena.de/romeo.htm). An artificial current dipole was placed at seven positions. Depth ranged from 43mm to 133mm with 15mm spacing. The dipole and electrode locations were digitized with a 3-D digitization system (3SPACE ISOTRAK II, Polhemus Inc., Colchester, VT, USA).

The phantom was first filled with a saline solution with the conductivity of 1.6 mS/cm. The surface potentials were recorded when activating each dipole with a sinusoidal 25-Hz current (amplitude 1 mA).

A layer of different conductivity was then inserted into the phantom. An artificial PVC-sternum was first mounted, with grooves on the left and right side. The thin inhomogeneity layer was fixed in these grooves. It was constructed of six 10´18 cm² acrylic frames. Ionic exchange membranes (NEOSEPTA type CM-2, NISSHO IWAI Deutschland GmbH, Duesseldorf, Germany) were fixed between these frames and PVC-covers. The membranes kept a steady concentration of NaCl in the compartments while the ionic current could flow freely. The maximal spacing between the inside wall of the phantom and the frames was 15 mm. The conductivity inside of the membrane was 16 mS/cm (i.e. 10 ´ the conductivity elsewhere in the phantom).

Fig. 1: a) Picture of phantom, dipole sources and the artificial inhomogeneity layer, respectively.

BEM computations

A homogeneous realistically shaped boundary-element model of the phantom was utilized. The torso surface was tessellated with 1252 nodes and 2500 triangular elements, and a constant conductivity of 0.16 S/m was assigned inside of it. A single moving equivalent current dipole (ECD) was used as the source model.

RESULTS

BSPM patterns

The measured potential distributions were presented as isocontour plots on the torso surface (Fig 2).

Fig. 2: Interpolated BSPM distributions on the surface of the triangulated torso model. a) The homogeneous phantom filled with 0.1 % solution. b) Phantom with an inhomogeneous sheet of 1 % solution.

Dipole localizations

The ECD localization results are summarized in Table I both for the homogeneous and inhomogeneous phantom.

TABLE I Dipole localization accuracies [mm] for the homogeneous and inhomogeneous phantom. P: dipole number, dx, dy, dz: localization errors in x, y, and z-directions, dr: absolute 3D-localization error.

	Homogeneous				1 % solution
P	dx	dy	dz	dr	dx	dy	dz	dr
1	-8.7	-0.4	0.9	8.8	-8.6	-6.4	16.8	20.0
2	-7.4	-0.7	-1.5	7.6	-14.2	-5.2	5.0	16.0
3	-7.8	-1.6	-4.1	9.0	-10.5	-5.4	0.5	11.9
4	-3.2	-1.6	-3.0	4.7	-10.8	-2.6	-1.6	11.3
5	-3.2	-1.2	-2.9	4.5	-4.3	-1.8	-1.6	5.0
6	-2.5	0.2	-5.2	5.8	-8.2	0.8	-5.1	9.7
7	-	-	-	-	-0.2	2.2	-5.2	5.6

DISCUSSION

The dipole localization results in the homogeneous phantom showed an excellent agreement with the digitized dipole locations (average 3D-error 6.7 mm). When the inhomogeneity layer was inserted the surface potential patterns did not exhibit significant changes (Fig. 2). The inhomogeneous maps had however slightly lower extrema amplitudes than corresponding homogeneous maps.

Average 3D error increased considerably (from 6.7mm to 11.3 mm) for the inhomogeneity layer. Its most significant impact was, however, on the localization accuracy of the dipole depth (dz). Dipoles close to the inhomogeneity layer seemed to be much deeper than the corresponding ones in the homogeneous phantom. The effect weakened for dipoles lying deeper (Table I).

In our previous finite-element (FEM) study with an anisotropic slab or sphere, unidirectional anisotropy between the dipole and the observation points stretched the potential map pattern in the direction of the highest conductivity [3]. The stretching of the map pattern increases the distance between the minimum and maximum of the field. Such a pattern could be falsely assumed to come from a source lying deeper than it actually is.

In this study, a fully anisotropic layer was not available. Still, the sheet of differing conductivity and the artificial sternum structure showed similar behavior for the dipole localizations than in the above mentioned FEM study: The dipole localizations were systematically too deep. More detailed FEM computations are still needed to assess more accurately how much error is introduced in the inverse solutions when the skeletal muscle layer and rib cage are not taken into account.

Acknowledgments: This work was financially supported by the Academy of Finland and Deutscher Akademischer Austausch Dienst.

REFERENCES

[1] K. Pesola, J. Nenonen, R. Fenici, J. Lötjönen, M. Mäkijärvi, P. Fenici, P. Korhonen, K. Lauerma, M. Valkonen, L. Toivonen, and T. Katila: Bioelectromagnetic localization of a pacing catheter in the heart. Phys Med Biol 44, 2565-2578, 1999.

[2] K. Simelius, I. Tierala, T. Jokiniemi, J. Nenonen, L. Toivonen, T. Katila: A body surface potential mapping system in clinical use. Med & Biol Eng & Comput 34(S1), 107-108, 1996.

[3] A. Pursula, J. Nenonen, E. Somersalo, R.J. Ilmoniemi, and T. Katila, “Bioelectromagnetic calculations in anisotropic volume conductors,” in Biomag2000, Proc. 12th Int. Conf. on Biomagnetism, J. Nenonen, R.J. Ilmoniemi, T. Katila, (Eds.), Helsinki Univ. of Technology, Espoo, Finland, 2001, pp. 659-662.