Analysis of the electromagnetic signals of the human brain: milestones, obstacles, and goals

Computerized mapping system of cerebral evoked potentials

The introduction of computerized analysis systems in the study of bioelectrical signals is enhancing the understanding of the physiological mechanisms which underlie cerebral evoked potentials (EPs) in response to externally applied stimuli. In the present study, short latency (0-50-ms) and long latency (0-500-ms) somatosensory evoked potentials (SEPs) were recorded by 32 scalp electrodes from normal and pathological subjects during median nerve stimulation. An interpolation procedure for estimating data values between the neighboring electrodes allowed the mapping of cortical activity across the scalp. Time signals were also transformed by an FFT algorithm and frequency maps obtained following the same interpolation procedure. Temporal and frequency maps were graphically displayed using color and three-dimensional plots. The usefulness of computerized topographical analysis is discussed; the time and frequency computer maps obtained from the same subjects are compared and their relative advantages are evaluated.

Real-time spectral intensity analysis of the EEG on a common microcomputer

An inexpensive system for quantitative electroencephalogram (EEG) analysis over long time periods is described. This system continuously analyses EEG for intensity versus frequency content, and saves averaged 'data reduced' intensity spectra to disk at regular intervals. The method had adequate frequency resolution for EEG analysis and demonstrated useful data reduction. In the example presented 1 day of sampled EEG data (20 megabytes) were reduced to 125 kilobytes of averaged spectra, thus illustrating the utility of the system for long term EEG studies. The system uses a low cost microcomputer (IBM AT compatible) with data acquisition board and a simple multitasking program developed with Asyst software.

Spatial patterns underlying population differences in the background EEG

A method is described which can be used to extract common spatial patterns underlying the EEGs from two human populations. These spatial patterns account, in the least-squares sense, maximally for the variance in the EEGs from one population and minimally for the variance in the other population and therefore would seem to be optimal for quantitatively discriminating between the individual EEGs in the two populations. By using this method, it is suggested that the problems associated with the more common approach to discriminating EEGs, significance probability mapping, can be avoided. The method is tested using EEGs from a population of normal subjects and using the EEGs from a population of patients with neurologic disorders. The results in most cases are excellent and the misclassification which occurs in some cases is attributed to the nonhomogeneity of the patient population particularly. The advantages of the method for feature selection, for automatically classifying the clinical EEG, and with respect to the reference-free nature of the selected features are discussed.

Beyond topographic mapping: towards functional-anatomical imaging with 124-channel EEGs and 3-D MRIs

A functional-anatomical brain scanner that has a temporal resolution of less than a hundred milliseconds is needed to measure the neural substrate of higher cognitive functions in healthy people and neurological and psychiatric patients. Electrophysiological techniques have the requisite temporal resolution but their potential spatial resolution has been not realized. Here we briefly review progress in increasing the spatial detail of scalp-recorded EEGs and in registering this functional information with anatomical models of a person's brain. We describe methods and systems for 124-channel EEGs and magnetic resonance image (MRI) modeling, and present first results of the integration of equivalent-dipole EEG models of somatosensory stimulation with 3-D MRI brain models.

New method for titrating differences in scalp topographic patterns in brain evoked potential mapping

A method is described for computing a Z estimator for the quantitative comparison of topographical patterns in 2 maps. Z can assume values between 1 and -1. The procedure is illustrated on a 3-shell head model for different configurations of current dipoles in the head space. The sensitivity of Z estimation can be adjusted by different weighting procedures or by using an average reference. When Z = 1 indicates identity of the set of active dipoles in the 2 maps compared, a dilation factor can be computed to estimate the enhanced or reduced activity of these generators.

Event-related covariances during a bimanual visuomotor task. I. Methods and analysis of stimulus- and response-locked data

A new method that measures between-channel, event-related covariances (ERCs) from scalp-recorded brain signals has been developed. The method was applied to recordings of 26 EEG channels from 7 right-handed men performing a bimanual visuomotor judgment task that required fine motor control. Covariance and time-delay measures were derived from pairs of filtered, laplacian-derived, averaged wave forms, which were enhanced by rejection of outlying trials, in intervals spanning event-related potential components. Stimulus- and response-locked ERC patterns were consistent with functional neuroanatomical models of visual stimulus processing and response execution. In early post-stimulus intervals, ERC patterns differed according to the physical properties of the stimulus; in later intervals, the patterns differed according to the subjective interpretation of the stimulus. The response-locked ERC patterns suggested 4 major cortical generators for the voluntary fine motor control required by the task: motor, somesthetic, premotor and/or supplementary motor, and prefrontal. This new method may thus be an advancement toward characterizing, both spatially and temporally, functional cortical networks in the human brain responsible for perception and action.

Dynamic functional topography of cognitive tasks

Improved neuroelectric recording and analysis tools are yielding increasingly specific information about the spatial and temporal features of neurocognitive processes. Such tools include recordings with up to 125 channels, digital signal processing techniques, and correlation of neuroelectric measures with anatomical information from magnetic resonance images. These tools, and their application to the study of cognitive functions, are presented in this paper.

The reference problem and mapping of coherence: a simulation study

Four techniques are applied to record EEG signals: bipolar recording, referential recording, common average reference recording and source derivation. For the interpretation of EEG parameter maps knowledge of the properties of the applied recording technique is essential. Bipolar recordings are not discussed in this paper. The application of reference and common average reference recording has the disadvantage of an unknown reference potential. This disadvantage is much larger with the use of source derivation because every electrode signal has its particular reference signal. This must be taken into consideration when coherence estimations are made. With actual EEG records the influence of the reference cannot be determined unambiguously. However, simulation studies enable some essential conclusions. In this paper by means of autoregressive processes EEG signals with given power and coherence properties were simulated and different recording situations using the same data set were reconstructed. The essential result is that computation and mapping of coherences yield the most reliable results when reference recording is used. However, measures to ensure a low reference signal must be taken.

Structural EEG analysis: an explorative study

A method is described to detect (subtle) changes in an EEG (electroencephalogram) by means of a Markovian modeling approach. This method, termed structural EEG analysis, treats the non-stationary EEG as a sequence of a finite number of short elementary patterns. Subtle changes in an EEG may be detected by studying the transition probabilities between the different patterns. By viewing the patterns as states in a Markov chain, a representation of the EEG structure based on a state transition probability matrix emerges. Various techniques to estimate the state transition probability matrices have been investigated. A number of experiments were performed with artificially generated data to determine the data length required to obtain a reliable estimate of the transition matrices. It appeared that a data length of approximately five to eight times the number of entries in the matrices is needed to accurately estimate the matrices. It was determined that the data length required to reliably estimate the transition probability matrix is dependent on the number of states and the number of non-zero entries of the matrix. Also, the data length appears independent of the values of the probabilities. The structural analysis approach was applied to actual EEG data, recorded from normal volunteers and epileptic subjects. It was demonstrated that visually confirmable changes in the EEG could be detected by the structural analysis method more accurately than by a more conventional approach.

Topographic mapping of the EEG: an examination of accuracy and precision

The accuracy and precision of topographic maps depicting scalp potentials and scalp potentials squared have been examined. Electrode placement was that specified by the International 10-20 System and the methods of interpolation bilinear and bicubic spines. The results indicate that, for these interpolation methods, the maximum error expected between the measured scalp quantities and those predicted by interpolation is positively correlated to the root-mean-square value of the measured quantity. Both interpolation methods produce accurate estimates of the interelectrode quantities. Both methods produce precise estimates of the scalp potential in the delta, theta and alpha frequency bands but only poor estimates in the beta band. The precision of the estimates of the scalp potentials squared is poor in all frequency bands. This result indicates that another look at the now common practice of topographically mapping the power-spectral components of the EEG is in order. In general, the bilinear and bicubic spline methods of interpolation perform about equally. This result is used to suggest that because of its additional computational complexity, use of the bicubic method for potential mapping may not be warranted. Advantages of the bicubic method, particularly in radial-current mapping, are however discussed.

Analysis of visual evoked potentials through Wiener filtering applied to a small number of sweeps

We introduce a method for processing visual evoked potentials, on the basis of a Wiener filter algorithm applied to a small number of consecutive responses. The transfer function of the filter is obtained by taking into account both the average of 99 sweeps (as an estimate of the true signal) and the EEG signal just before the stimulus onset (as an estimate of the noise superimposed on each individual response). The process acts as a sweep-by-sweep filter (in the sense of the mean square error) which considers the possible non-stationarities of the EEG signal during a complete clinical procedure. The average of a small number of consecutive filtered sweeps reveals variations in the morphology of the evoked responses which produce a change in the principal latencies. Applications are foreseen in neurophysiological studies of visual evoked potential responses, and in the clinic, where it is important to evaluate adaptive mechanisms, dynamic changes in single groups of visual evoked potentials and cognitive responses.

Mapping of scalp potentials by surface spline interpolation

Evoked potentials and EEGs record punctate electrical activity at electrode sites. To represent the overall potential distribution on the entire scalp it is necessary to interpolate between these sampled values. Surface splines are mathematical tools for interpolating functions of two variables. In comparison to the classical methods of interpolation, based on linear combination of the potentials of the 4 nearest electrodes, spline methods are smoother, give more precisely located extrema and converge faster toward the 'true' potential surface when the number of recording electrodes is increased. These advantages are at the expense of lengthier computation time.

Improved event-related potential estimation using statistical pattern classification

A new method of ERP estimation with minimal statistical assumptions is presented. A mathematical pattern classification procedure is used to select trials with discriminable event-related signals in a time interval of interest. A method of forming a reference 'baseline' is also presented. Stimulus-registered and response-registered 'enhanced' ERP averages computed from selected trials of a visuo-motor experiment show substantial enhancement of event-related signals, especially for channels with weak signals, while rejected trials have minimal event-related signals.

Classification of the EEG during neurosurgery. Parametric identification and Kalman filtering compared

A procedure is described which aims to classify an EEG recorded during neurosurgery, for example intracerebral aneurysm clipping. A parametric approach is used; it employs auto-regressive (AR) modelling and Kalman filtering to quantify directly the dynamics of the EEG generating mechanism, supposing it to be a linear, time-invariant system driven by white noise. The results of this EEG processing are analysed together with simultaneous values of arterial blood pressure (ABP) as surgery of this kind is carried out under conditions of controlled hypotension. The object is to compare the sensitivity of ABP data with that obtained from the EEG and so provide an early warning of a potentially dangerous non-physiological state induced by the hypotensive drug (in this case sodium nitroprusside). Some methodological comments on the correct implementation of these algorithms are given and the procedure is compared with similar approaches which have appeared in the literature during the last few years. Particular emphasis is placed on the power spectral analysis of the signal by pointing out a method for spectral decomposition, related to AR power density estimation, which permits the separation of single spectral components in terms of central frequencies and their associated power. Other potential applications of this method are in long term EEG monitoring for the detection of changes due for example to drug infusion, to fast transient events, or to changes in the stationary condition.

The brain's magnetic field: some effects of multiple sources on localization methods

The theoretical basis for magnetic field recording (MEG) methods is briefly summarized. Lines of constant radial magnetic field on a spherical surface, which are typically used in MEG applications to locate sources, are shown for various multiple dipole sources. It is shown that the usual localization methods are subject to relatively large error if only one additional dipole is present. New methods to improve spatial resolution are discussed.