The success of surgical intervention in the partial epilepsies is crucially affected by the accuracy of pre- and intraoperative source location techniques. Several approaches to the localization problem have been employed, that with the longest history being scalp-recorded EEG. Despite considerable advances in other imaging technologies such as MRI and PET, localization via the electrical signals generated by epileptic brain continues to provide the data most relied upon in pre-operative assessment. The present paper presents an overview of the contribution of various localization techniques. It is argued that electrical signals of the brain, as represented by EEG and MEG, remain the best methods to locate sources, and that the application of analysis techniques presently under investigation will further improve the accuracy of the non-invasive scalp-EEG approach.

The success of surgical intervention in the partial epilepsies is crucially affected by the accuracy of pre- and intraoperative source location techniques. Several approaches to the localization problem have been employed, that with the longest history being scalp-recorded EEC Despite considerable advances in other imaging technologies such as MRI and PET, localization via the electrical signals generated by epileptic brain continues to provide the data most relied upon in pre-operative assessment. The present paper presents an overview of the contribution of various localization techniques. It is argued that electrical signals of the brain, as represented by EEG and MEG, remain the best methods to locate sources, and that the application of analysis techniques presently under investigation will further improve the accuracy of the non-invasive scalp-EEG approach.

Theoretical physics predicts that currents in biological media result in magnetic fields; however, the predicted fields are so small that they were measurable only recently. One of the first biomagnetic measurements was of fields associated with heart function, measured by Cohen et al. in 1970. Coil magnetometers of the kind they used are usually not sensitive enough for the detection of brain function, which is an order of 10−4 smaller than fields produced by the heart. Consequently it was not until a Josephson junction was incorporated into a superconductive quantum interference device (SQUID) that magnetometers with the required high sensitivity were available for the measurement of brain function. The SQUID is used as ultrasensitive magnetic flux detector. The problem with SQUIDs is that, in order to maintain their superconductivity, the sensor has to be cooled to the temperature of liquid helium (4.2° K). In order to accomplish these low temperatures the SQUID is immersed in liquid helium inside a helium dewar. The rf-SQUID is a superconducting ring with one Josephson junction (weak link) in it. The do-SQUID has two weak links in the ring. Flux transformers transfer flux from a sensing coil to the SQUID. For example, a closed loop of superconducting wire maintains the total magnetic flux inside the loop. If this loop contains two coils, coupled in series, a change of the magnetic flux through one of the coils causes a change in the magnetic flux in the other coil. Thus, magnetic flux is transferred from the sensing coil Ll to the signal coil Ls inside the SQUID (Figure 25.1). In order to increase the signal-to-noise ratio, differential magnetometers, referred to as gradiometers, are utilized in preference to the simple magnetometer (Figure 25.2). The first-order gradiometer has two sensing coils, Ll and L2.

Although a great deal has been learned from the study of scalp-recorded movement-related potentials, the exact origin of the various components described above remains unclear, mainly because of the limited spatial resolution of the EEG and the consequent difficulty in predicting sources from the surface distribution of these components. The introduction of the MEG and source localization methods based on neuromagnetic recordings has provided a new means by which to study the cortical activation during movement in humans. The study of movement-related magnetic fields of the brain is still at an early stage of development. However, the data reviewed to date indicate that the MEG offers a promising means by which to study (noninvasively) cortical motor function. With regard to the data reviewed here, the following conclusions can be made: 1. Slow "readiness" magnetic fields can be recorded prior to a variety of voluntary movements and display a topography which indicates the activation of bilateral sources, even if the instructed movement is unilateral. Sources in the contralateral hemisphere appear as early as 0.5 sec before the movement and appear to be localized in the sensorimotor cortex. Consequently, the assumption of a contralateral source being the only or primary generator of the readiness potential, based on EEG data, must be tempered. 2. A large-amplitude "movement-evoked field" (occurring at a post-EMG-onset latency of about 110 msec for finger movements) is probably the counterpart of the MP and appears to be the result of a dipolar source localized to the contralateral sensorimotor area. This source is probably the first sign of movement reafferent input to cortex. 3. Variability in the movement-evoked field across individuals, which are much more evident in MEG than in EEG, may reflect the summation of multiple sources active in the region of the sensorimotor cortex during movement onset (i.e., both pre- and postcentral generators). In some instances, it may be possible to extract simpler elements of these complex sources based on assumptions of temporal overlapping of pre-movement and movement-evoked activity.

The clinical literature has suggested that while the clinical features and presentation of benign rolandic epilepsy in children (BREC) are known, the neuronal mechanism of the epileptic focus is poorly understood. Classification of clinical subtypes is usually made by determining whether there are supplementary clinical signs of brain damage, in which case the epilepsy is classified as non-benign or "atypical". Studies of EEG findings in BREC have suggested that the source of the epilepsy is in the Rolandic fissure. We investigated dipole source modelling in 24 children, comparing the results of one and two dipole models. The results indicate that atypical BREC patients have a more complex distribution of dipoles and that single dipole fits may be more predictive of typical BREC than multiple dipole fits. The implications of these results are discussed.

This study examined the effect of modality of stimulation on two measures of cerebral function: (a) the scalp distribution of sensory evoked potentials and (b) the cerebral distribution of radiolabelled HMPAO. Steady-state stimulation in the auditory, somatosensory or visual modality was presented to six subjects. Scalp potentials were measured from a distribution of electrodes, and the radiopharmaceutical was injected through an indwelling intravenous catheter midway through the stimulation/ recording session. Equivalent dipole sources estimated from the spatial distribution of the scalp potentials were found to be consistent with regions of high HMPAO uptake as imaged by Single Photon Emission Computed Tomography (SPECT).

Abstract : Stereopsis, is an important and interesting subject of study for many reasons. Clearly, the brain codes, and stores, complex information about the relative positions of objects in space; how this is accomplished is not only important for an understanding of stereopsis but could also be instructive for the analysis of other processing systems in the brain. For example, disparate visual input is simultaneous input and, from what is known about stereopsis, that processing appears to occur in parallel. Local stereopsis is the point-by-point comparison of the images on the two retinae and the calculation of the associated disparity for every feature on the visual image. Global stereopsis is the integration of all the coded, local disparities into an overall perception of a three dimensional scene. (js)

The application of magnetoencephalography (MEG) to the analysis of sources in the brain responsible for early and late components of evoked potentials is discussed. Representative data are presented and discussed which demonstrate localisation of sources assuming single equivalent dipoles. Distributed systems as sources for some steady-state responses are discussed in relation to the broader issue of the usefulness of equivalent single dipole models. These issues are related to the use of MEG for the analysis of source systems influenced by alcohol and other drugs.

A method is described which estimates the location of current dipoles in the brain. The estimate establishes the best fit, according to stated criteria, between observed magnetic fields measured with a third-order gradiometer and fields predicted from one or two dipoles. The method takes into account corrections for the number, spatial separation and size of the gradiometer coils, and for the inclination of the axis of the gradiometer to the radius from a computed origin in the head to the centre of the sensing (first) coil, or to the position of observation on the head. The method also provides for a means to establish the best initial guess for a dipole location and contains procedures for estimating the positions of more than two dipoles in appropriate cases. Several relevant theoretical considerations are discussed and a number of applications of the method are described.