Comprehensive spectral analysis of human EEG generators in posterior cerebral regions

Cross-Frequency Brain Network Dynamics Support Pitch Change Detection

Processing auditory sequences involves multiple brain networks and is crucial to complex perception associated with music appreciation and speech comprehension. We used time-resolved cortical imaging in a pitch change detection task to detail the underlying nature of human brain network activity, at the rapid time scales of neurophysiology. In response to tone sequence presentation to the participants, we observed slow inter-regional signaling at the pace of tone presentations (2-4 Hz) that was directed from auditory cortex toward both inferior frontal and motor cortices. Symmetrically, motor cortex manifested directed influence onto auditory and inferior frontal cortices via bursts of faster (15-35 Hz) activity. These bursts occurred precisely at the expected latencies of each tone in a sequence. This expression of interdependency between slow/fast neurophysiological activity yielded a form of local cross-frequency phase-amplitude coupling in auditory cortex, which strength varied dynamically and peaked when pitch changes were anticipated. We clarified the mechanistic relevance of these observations in relation to behavior by including a group of individuals afflicted by congenital amusia, as a model of altered function in processing sound sequences. In amusia, we found a depression of inter-regional slow signaling toward motor and inferior frontal cortices, and a chronic overexpression of slow/fast phase-amplitude coupling in auditory cortex. These observations are compatible with a misalignment between the respective neurophysiological mechanisms of stimulus encoding and internal predictive signaling, which was absent in controls. In summary, our study provides a functional and mechanistic account of neurophysiological activity for predictive, sequential timing of auditory inputs.SIGNIFICANCE STATEMENT Auditory sequences are processed by extensive brain networks, involving multiple systems. In particular, fronto-temporal brain connections participate in the encoding of sequential auditory events, but so far, their study was limited to static depictions. This study details the nature of oscillatory brain activity involved in these inter-regional interactions in human participants. It demonstrates how directed, polyrhythmic oscillatory interactions between auditory and motor cortical regions provide a functional account for predictive timing of incoming items in an auditory sequence. In addition, we show the functional relevance of these observations in relation to behavior, with data from both normal hearing participants and a rare cohort of individuals afflicted by congenital amusia, which we considered here as a model of altered function in processing sound sequences.

Interaction between changes in local and temporospatial spectral EEG characteristics during exposure of humans to hypoxia

The temporospatial and local characteristics of the EEG were studied in healthy subjects during the respiration of a hypoxic oxygen-nitrogen gas mixture containing 8% oxygen. Analysis of the spectral power density, coherence, phase shift, similarity of dominant frequencies in the EEG in different leads was performed separately for epochs containing and not containing visually apparent patterns of EEG spatial synchronization. In addition, assessment of spectral measures took account of the fact of the dominance of the frequency being analyzed in the EEG spectrum of the corresponding lead. The results of these studies showed that overall, hypoxia was accompanied by a decrease in EEG coherence in the alpha range and, for connections in the left mid-temporal lead, the delta range. The beta range showed an increase in the values of this measure, while the theta range showed different changes in coherence: increases in epochs containing spatial synchronization patterns and decreases from control levels in other epochs. Hypoxia was also associated with increases in the EEG phase shift in the frontal and temporal leads (in relation to the EEG recordings in the other leads) in the delta and theta ranges. In the beta range, conversely, there was a decrease in the mean phase shift. Consideration of the fact of dominance of one or another frequency in the spectra of local EEG traces is needed for correct interpretation of analyses of temporospatial measures. A mathematical model of the interaction between processes at different frequencies is presented to explain some of the observations in the current work.

Nonlinear multivariate analysis of neurophysiological signals

Multivariate time series analysis is extensively used in neurophysiology with the aim of studying the relationship between simultaneously recorded signals. Recently, advances on information theory and nonlinear dynamical systems theory have allowed the study of various types of synchronization from time series. In this work, we first describe the multivariate linear methods most commonly used in neurophysiology and show that they can be extended to assess the existence of nonlinear interdependence between signals. We then review the concepts of entropy and mutual information followed by a detailed description of nonlinear methods based on the concepts of phase synchronization, generalized synchronization and event synchronization. In all cases, we show how to apply these methods to study different kinds of neurophysiological data. Finally, we illustrate the use of multivariate surrogate data test for the assessment of the strength (strong or weak) and the type (linear or nonlinear) of interdependence between neurophysiological signals.

Assessing the accuracy of topographic EEG mapping for determining local brain function

OBJECTIVE

There has been considerable discussion regarding the accuracy of topographic electroencephalographic (EEG) maps for assessing local cerebral function. We performed this study to test the accuracy of EEG mapping by examining the association between electrical activity and the perfusion under each electrode as another measure of local cerebral function.

METHODS

EEG mapping was performed simultaneously with (H15)2O positron emission tomography (PET) scanning in 6 normal adult subjects, both at rest and during a simple motor task. EEG data were processed using 3 different montages; two EEG power measures (absolute and relative power) were examined.

RESULTS

Relative power had much stronger associations with perfusion than did absolute power. In addition, calculating power for bipolar electrode pairs and averaging power over electrode pairs sharing a common electrode yielded stronger associations with perfusion than data from referential or single source montages.

CONCLUSIONS

These findings indicate (1) that topographic EEG mapping can accurately reflect local brain function in a way that is comparable to other methods, and (2) that the choice of EEG measure and montage have a significant influence on the degree with which maps reflect this local activity and function.

The dynamics of phase relationships of alpha waves during photic driving

Inter- and intrahemispheric phase relationships of alpha waves during photic driving with intermittent flash stimuli were successively calculated by our newly developed method and compared with the results obtained during the resting condition. Of 12 healthy subjects, 6 showed obvious photic driving responses in the occipital regions (driven group) but the remaining 6 did not (non-driven group). In the driven group, the interhemispheric phase differences decreased, especially when EEG frequencies coincided with stimulating frequencies. However, this was not maintained throughout the stimulation period but was accompanied by some fluctuation. The interhemispheric phase difference decreased in the posterior temporal and, in particular, in the occipital areas. The intrahemispheric phase difference for the occipito-frontal and the occipito-central pairs in each hemisphere tended to increase and remained at approximately pi radians, especially for the former pair. These findings suggest the presence of a specific response in the occipital area. Time ratios of the phase lead in longitudinal and transverse directions and the high similarity of the time courses of intrahemispheric phase differences between bilaterally homologous pairs at rest suggest that there exist phasing mechanisms which act in parallel between the left and right hemispheres with dominant antero-posterior time delay.

Half-field sinusoidally modulated light stimulation at subject's alpha frequency

Interhemispheric responses of alpha activity were investigated by use of half-field sinusoidally modulated light (SML) stimulation at the subject's alpha frequency. The left and right visual half-fields as well as full-field were separately stimulated by the SML. The power spectrum of SML responses, averaged with Wiener filtering, was obtained at the left and the right occipital area (O1 and O2) in 11 normal subjects. Power in each hemisphere, phase difference and coherence between O1 and O2 were estimated at an alpha frequency. A laterality index of power was defined as (P(l)-P(r))/(P(l) + P(r)) where P(l) is a power value at O1, and P(r) at O2. In most of the subjects, laterality index decreased in the order: left half-field, full-field and right half-field stimulation. This relationship revealed greater SML responses on the ipsilateral occipital area. There were significant differences in laterality index among 3 visual field conditions, thus showing that full-field responses ranged in lateralization between the left and right half-field responses in most of the subjects. This hemispheric distribution suggests that half-field SML stimulation affected alpha activity selectively in each hemisphere, mostly in the ipsilateral hemisphere.

Generation of human EEG by a combination of long and short range neocortical interactions

A theory of neocortical interactions is developed involving both local delays (PSP rise and decay times) and global delays due to finite velocity of action potentials in corticocortical fibers. The theory is based on plausible assumptions regarding input/output relations in neocortical columns and realistic neural parameters. The simultaneous existence of short wavelength waves propagating away from multiple epicenters and long wavelength standing waves due to global boundary conditions is predicted. Both phenomena appear to have dominant oscillation frequencies in the general range of observed EEG phenomena in humans. A mechanism by which removal of diffuse input from the reticular formation may cause an abrupt drop in EEG frequency (as in the transition from the awake to sleeping state) is postulated.

Transverse relationships of the alpha rhythm on the scalp

Using auto- and cross-spectral analyses on data of 3 normal adult subjects, two components of rhythmic alpha activity were distinguished with respect to their spatial distribution and spatial relationships on the scalp. Derivations from 21 scalp points vs. linked ears were analyzed. The alpha activity with power maxima in the bilateral occipital areas (component A) showed high coherence with activity in the anterior areas and somewhat lower coherence with activity in the central/parietal areas. The other component (B) appeared dominantly in the central areas, showing extremely low values of coherence between the anterior and posterior regions, and an unstable phase relation among recording points. Components A and B appear to correspond to the 'generalized' and 'localized' alpha components described earlier. It is suggested that these two components account for the major properties of the rhythmic alpha activity.

Spatial distribution of generators of alpha activity

Source determination of alpha activity was studied using the relative power contribution analysis (RPCA) method which allows determination of the relative contributions of different areas to the power of a certain area at different frequencies. In 20 normal subjects, EEGs were recorded from F3, F4, C3, C4, P3, P4, O1 and O2, each referenced to a linked ear. An 8-dimensional autoregressive model was fitted to the EEGs of 10.24 sec. Based on the model, RPCA was performed. For each area, alpha activity was divided into two parts: one originating in its own area (endogenous) and another in the other areas (exogenous). Endogenous alpha activity increased as the area was more posterior. In the anterior regions (frontal and central), endogenous alpha power (power of endogenous alpha activity) was small, while exogenous alpha power was large. In the posterior regions (parietal and occipital), the amount of endogenous alpha power did not differ markedly from that of exogenous alpha power. The posterior regions, which generate more endogenous alpha activity, can be considered to play a dominant role in alpha generating mechanisms. In some subjects, alpha generators with a different frequency from that of the occipital areas were observed.

The direction of spread of alpha activity over the scalp

The direction of spread of alpha activity over the scalp was studied using the modified 'entropy analysis' method which distinguishes the direction of time differences between two correlated signals in the alpha band. The resting EEG was recorded from F3, F4, C3, C4, P3, P4, O1 and O2, each referenced to linked ears in 10 subjects. The directed measure of correlation of alpha activity in two directions (e.g., from O1 and O2 and from O2 to O1) was obtained from all possible pairs of electrodes. The dominant longitudinal direction of alpha activity was found to be anterior to posterior within as well as across hemispheres. No dominant transverse direction was found. This suggests that the spread of alpha activity over the scalp is in the dominant anterior-posterior direction.

Inter- and intrahemispheric relationships of the EEG activity during sleep in man

Using both linear and non-linear multivariate analysis associated with classical statistical tests, inter- and intrahemispheric relations of the cerebral electrical activity (EEG) during sleep were analysed by spectral power, coherence (CH) and correlation functions. Whilst a steady diffuse decrease in linear phase relations (CH) appeared during the evolution from waking to slow wave sleep, paradoxical sleep was associated with low interhemispheric CHs (mostly frontal) and high intrahemispheric CHs. Simultaneously, at a much lower degree of temporal resolution (25.6 sec periods), the correlations between EEG activities of identical frequencies but of different topographies were highly positive during sleep, whilst the correlations between activities of different frequencies at the same site on the scalp varied according to sleep states and frequencies. These findings suggest the interpretation of deep sleep in man as a state of transient partial disconnection between hemispheres.

Methods for finding single generators, with application to auditory driving of the human EEG by complex stimuli

A method is presented for searching any number of simultaneous time series (e.g. from different EEG electrodes) for single generators within a frequency band. The method was applied to human EEGs taken during conditions of listening to complex rhythmic (Mozart symphony and chanting) and non-rhythmic (conversation) sounds. The method indicated a tendency towards single generation in the alpha (8-12 Hz) band during the rhythmic conditions. The method was then extended to test for synchrony between the EEG signals and the auditory signal (as a function of frequency). Such synchrony (auditory driving) was indicated during rhythmic conditions in the alpha band.

EEG changes induced by light stimuli modulated with the subjects alpha rhythm

A new method, named the 'alpha wave modulated light' (AML) stimulation, has been used for controlling the frequency of sine wave modulated light by the subject's on-going alpha rhythm. The EEG was recorded from O1 and O2. With the AML stimulation using either the left occipital alpha rhythm or the right, two phase differences of O degrees and 180 degrees were created between AML and the alpha rhythmn. The peak frequency and power of occipital alpha in both hemispheres and the coherence and phase between the left and right occipital EEG were obtaned for 7 subjects. The alpha power was larger at a 180 degrees phase shift than at a 0 degrees phase shift with ipsilateral AML stimulation. The ipsilateral increase in the alpha power was greater than the contralateral at a 180 degrees phase shift. Larger coherence value and right phase lead were observed at 180 degrees phase shift by use of the right alpha rhythm. The possible mechanisms of influences of AML stimulation on the alpha activity were discussed.

A comparison of EEG activity in the left and right cerebral hemispheres by power-spectrum analysis during language and non-language tasks

In 10 female subjects, power-spectrum analysis was performed on the alpha activity elicted during the resting state, 4 right hemispheric tasks, and 3 left hemispheric tasks. The data were treated in 3 ways: approach 1, comparing the right and left hemispheric alpha activity; approach 2, comparing the right and left hemispheric alpha activity adjusted for the resting state; and approach 3, comparing the right and left hemispheric alpha activity adjusted for the previous task. Approaches 1 and 2 revealed few significant differences in the alpha activity of the hemispheres, but approach 3 provided data that better fit the theory of decreased power spectrum of attenuation of activity in the activated hemisphere. Thus, approach 3 may be useful in developing an electroencephalographic test for determining cerebral dominance for language.

Interhemispheric relationships of reponses to sine wave modulated light in normal subjects and patients

(1) The interhemispheric amplitude correlations of the fundamental (A1) and second harmonic (A2) components of responses to sine wave modulated light were determined in the occipital, parietal and temporal scalp areas of a group of normal subjects and a group of patients with unilateral irritative EEG phenomena. The interhemispheric amplitude correlation was computed at 10- and 16-c/sec stimulation frequencies and under 3 conditions: no modulation and no attention (M-, A-), modulation 30%, and no attention (M+, A-) and modulation 30% and attention (M+, A+). The correlations wre expressed in the rank correlation coefficient of Kendall (rK). (2) Analysis of variance revealed that the interhemispheric amplitude correlation of A1 in the normal group was significantly larger than that in the patient group at a stimulation frequency of 10 c/sec and under the conditions (M-, A-) and (M+, A-). (3) Both groups presented the largest correlation coefficients in the occipital scalp area. (4) In the normal group a significant increase of rK occurred under the influence of modulation. Attention caused a decrease of the correlation coefficient of A1. In the group of patients this influence was not clear. (5) The standard deviation of the mean interhemispheric phase differences was considered as a measure of synchrony of the response components between the two hemispheres. The smallest values of standard deviation were found in the occipital scalp area of the normal group at 10 c/sec stimulation frequency. Considerable variations between individuals were observed in both groups. (6) The correlation between the interhemispheric amplitude correlation (rK) and the interhemispheric synchrony was influenced by changing modulation depth from 0 to 30% and by introduction of attention, mostly in the group of normal subjects. This phenomenon was considered to support the theory that in patients with unilateral EEG disturbances the interhemispheric relations are different from those in normal subjects. (7) The findings as to amplitude correlation (rK) were the same under the condition (M-, A-) as under the condition (M+, A-).

The relationship of head size to alpha frequency with implications to a brain wave model

Analogies between brain waves and waves in physical systems suggest that EEG frequency may be partly determined by cortical surface area. Since a number of other physiological and anatomical parameters probably influence EEG frequency, only a weak correlation is to be expected. A study was made of 159 subjects, some of whom had either very large or very small heads. A single number representing head size was determined as the cube root of three linear measurements. Several characteristic EEG frequencies were determined for each subject by means of Fourier analysis. The data indicate that alpha frequency is significantly correlated with head size: larger heads tend to produce slower alpha rhythms. It was also shown that alpha frequency tends to be lower in all subjects above roughly age 60. Subjects above produced significantly less alpha rhythm than the younger group. It is suggested that analogies between brain waves and physical waves may explain a number of phenomena which are typical of EEG.

A real-time software system on the PDP-11 for two-channel EEG spectral analysis during surgery

Three functions are performed in real time by a stand-alone PDP-11-assembly language program. First two channels of EEG data are acquired by quantization techniques; second spectral analysis of the EEG data is carried out; third the resulting spectra are displayed as time compressed spectral profiles on an X-Y plotter. The method provides EEG time trends during anesthesia, permitting evaluation of brain functions.

The localization of equivalent dipoles of EEG sources by the application of electrical field theory

A technique is described for finding the position, magnitude and orientation of the equivalent electrical dipole of EEG activity given the pattern of potential differences recorded at the scalp. The technique is based on an iterative computer program implementing equations describing the electrical field of a dipole in a spherical conductor. The computer program was tested in vitro against data obtained from an inert spherical conductor (a bowl containing physiological saline, fitted with recording electrodes and a movable dipole) and an anisotropic conductor (a similarly equipped human skill including a simulated scalp). In both practical models, the computer program accurately located the dipole, the mean differences between observed and computed loci being of the order of 1 cm at widely different locations. This accuracy was maintained in the anisotropic model even though potentials transmitted through the skill were attenuated by 80%. In vivo, the program successfully located the equivalent generator of blink artefacts within the remaining eye of a one-eyed subject and, in a normal subject, localized the dipole of corresponding potentials to a midline inter-ocular position. In further investigation of normal subjects, the distribution of amplitudes at latencies within Wave V of the visual evoked response confirmed the loci of equivalent generators within posterior cerebral regions. The technique was also applied to the alpha rhythm indicating a posterior locus compatible with the view that alpha rhythm is generated chiefly by posterior cerebral cortex. Factors affecting the accuracy of the technique and the limitations of one-dipole models are discussed.

Spectral analysis of the EEG recorded during stimulation of the human fovea

Computer spectral analysis was performed on the EEGs of 6 subjects recorded during monocular foveal stimulation. Stimulus wavelength and subjective brightness were varied independently to determine their importance to changes in the ongoing EEG. The contribution of oculomotor control to these effects and the retinal area of their origin were also investigated. The major results were as follows. (1) Stimulation of the fovea reduced the amplitude of the EEG and the coherence between hemispheres at all frequencies, but most dramatically in the alpha band. (2) It increased the variance of EEG amplitudes and widths of spectral peaks in the alpha band and shifted these peaks to lower frequencies. (3) Auto-spectral intensities in the alpha band were enhanced slightly at brightness levels near photopic threshold, but were unaffected by changes in stimulus wavelength. (4) The fovea appeared to be the most effective retinal area in which irradiation attenuated ongoing alpha activity.

Bispectrum analysis of electroencephalogram signals during waking and sleeping

The degree of interaction of component waves making up a single electroencephalogram trace was strongly correlated with alpha activity, lead placement, and state of consciousness. Significant quadratic coupling of the waves was found only for awake subjects with high alpha activity. For these subjects about 50 percent of beta activity can be attributed to harmonic coupling with the alpha peak. During sleep, the degree of interaction was of borderline significance and did not follow a consistent pattern with respect to subject, frequency, state, or lead.

Gaussian behavior of the electroencephalogram: changes during performance of mental task

The probability distribution of the amplitude of scalp electroencephalogram has been investigated in an adult subject in the idle state, and during performance of a mental arithmetic task. Based on a large sample, the electroencephalogram in this subject in the idle state follows a Gaussian (normal) probability function 66 percent of the time. During performance of the arithmetic task, the portion of Gaussian electroencephalogram decreases to 32 percent. The probability function characterizing gross electroencephalographic activity is determined by the degree of mutual interaction of individual cellular generators of wave activity in the tissue underneath the recording electrode. The data imply an increase in the cooperative activity of cortical neuronal elements during performance of a mental task.