Marker guided registration of electromagnetic dipole data with tomographic images

The registration method proposed in this paper is much more accurate than the present-day source estimations, and hence will keep its value when improved dipole models are developed.

Abstract

Anatomical interpretation of EEG and MEG derived source estimations is difficult, and their correspondence with pathology as revealed by medical images is hard to assess. In this study a method is presented to register electromagnetic source data with tomographic image data of the same patient, thus facilitating the interpretation of the dipole characteristics with respect to the patient's anatomy (MRI, CT) or metabolism (SPECT). The method utilizes external triangular markers that are easy to apply to the skin and indicate reference points with subslice accuracy, even if these are located slightly outside the scanned volume. In this way accurate matching is ensured not only in high resolution images but also in standard CT and MR imaging protocols employing thick slices and/or large interslice gaps. While a similar triangular marker can be used for SPECT imaging, point-like radioactive markers have been considered as well because of their simplicity. At present no final conclusions can be drawn about the optimal design of the SPECT marker. The clinical potential of dipole source modelling in epilepsy and other neurological applications has not yet been established, mainly because the accuracy of the source estimations is still uncertain. The registration method proposed in this paper is much more accurate than the present-day source estimations, and hence will keep its value when improved dipole models are developed.