T. Katila

Vortex shaped current sources in a physical torso phantom.

Recent studies reported differential information in human magnetocardiogram and in electrocardiogram. Vortex currents have been discussed as a possible source of this divergence. With the help of physical phantom experiments, we quantified the influence of active vortex currents on the strength of electric and magnetic signals, and we tested the ability of standard source localization algorithms to reconstruct vortex currents. The active vortex currents were modeled by a set of twelve single current dipoles arranged in a circle and mounted inside a phantom that resembles a human torso. Magnetic and electric data were recorded simultaneously while the dipoles were switched on stepwise one after the other. The magnetic signal strength increased continuously for an increasing number of dipoles switched on. The electric signal strength increased up to a semicircle and decreased thereafter. Source reconstruction with unconstrained focal source models performed well for a single dipole only (less than 3-mm localization error). Minimum norm source reconstruction yielded reasonable results only for a few of the dipole configurations. In conclusion active vortex currents might explain, at least in part, the difference between magnetically and electrically acquired data, but improved source models are required for their reconstruction.

Cardiomagnetic Source Imaging Studies with Focal and Distributed Source Models

During the recent years, niagnetocardiographic (MCG) source imaging has received increasing iuterest [l, 2). The ability to locatc currcnt sourccs combincd with prccisc timing of cvcnts is valuable both in basic rcscarch and in clinical studies [3, 4]. Prcsc.ntly, thc MCG is employed at sorae hospitals to test and further develop its clinical use [5, 6|. The BioMag Laboratory at the Helsinki University Central Hospital (HUCH) is equipped with gtate-ofthe-art facilities both for basic research and for clinical applications in functional ncuro and cardiomagnctic studics. Thc cardiac mcasurcmcnt Systems tLc 67clianncl MCG and thc 123-channcl body-surfacc potential mapping (BSPM) equipment have been described in detail elsewhere [7). Combiued with fast data transfer to the Helsinki University of Technology (HUT) and the Signal and image processing methods developed at HUT, the facilities give a good iramework for cardiomagnetic source imaging research. In the following, we describe briefly such studies in association with the ongoing patient studies. First, we outline howindividualized boundary-eleinent torso models are constructed froni inedical image data. We have developed nearly automated methods to segmcnt thc objccts of intorcst from magnctic rcsonancc (MR) or X-ray imagcs, äs well äs a novcl triangulation Software to tessellate the surfaces of the objects. Next, we discuss testing of the ability of MCG to locate point-like sources. Artificial current-dipole sources have been utilized. both in phantom experiinents and in studies with a pacing catheter placed inside the heart during the MCG recordings. In addition. we show rcsults obtaincd with distributcd source modcls, such äs minimum-norm cstimatcs on thc cpicardial surface, and double-layer sources in reconstructing activation sequences on the heart surfaces. In our studies, the golden Standard for the inverse Solutions is provided by clinically validated invasive data recorded before MCG measurements. In many cases, the BSPM was performed simultaneously with thc MCG rccordings, and similar source analysis was carricd out both from thc MCG and thc BSPM data. Methods Patient studies