Dietrich Lehmann

Low resolution electromagnetic tomography: a new method for localizing electrical activity in the brain

This paper presents a new method for localizing the electric activity in the brain based on multichannel surface EEG recordings. In contrast to the models presented up to now the new method does not assume a limited number of dipolar point sources nor a distribution on a given known surface, but directly computes a current distribution throughout the full brain volume. In order to find a unique solution for the 3-dimensional distribution among the infinite set of different possible solutions, the method assumes that neighboring neurons are simultaneously and synchronously activated. The basic assumption rests on evidence from single cell recordings in the brain that demonstrates strong synchronization of adjacent neurons. In view of this physiological consideration the computational task is to select the smoothest of all possible 3-dimensional current distributions, a task that is a common procedure in generalized signal processing. The result is a true 3-dimensional tomography with the characteristic that localization is preserved with a certain amount of dispersion, i.e., it has a relatively low spatial resolution. The new method, which we call Low Resolution Electromagnetic Tomography (LORETA) is illustrated with two different sets of evoked potential data, the first showing the tomography of the P100 component to checkerboard stimulation of the left, right, upper and lower hemiretina, and the second showing the results for the auditory N100 component and the two cognitive components CNV and P300. A direct comparison of the tomography results with those obtained from fitting one and two dipoles illustrates that the new method provides physiologically meaningful results while dipolar solutions fail in many situations. In the case of the cognitive components, the method offers new hypotheses on the location of higher cognitive functions in the brain.

Sleep deprivation: effect on sleep stages and EEG power density in man.

Sleep was analysed in 8 young adults subjects during two baseline nights and two recovery nights following 40.5 h sleep deprivation. Sleep stages were scored from the polygraph records according to conventional criteria. In addition, the EEG records of the entire nights were subjected to spectral analysis to compute the frequency distribution of the power density in the 0.25-25 Hz range for 0.5 Hz or 1.0 Hz bins. In the first recovery night, the power density in the delta band was significantly higher than baseline for total sleep time as well as for sleep stages 2, 3 and 4, 4 and REM. These changes were not restricted to the delta band, but extended to higher frequency bands. Minor, but significant, effects of sleep deprivation were seen in the power density distribution of the second recovery night. In the baseline nights, a progressive reduction of power density in the delta/theta range was present for successive non-REM-REM sleep cycles for total sleep time and stages 2, 3 and 4, and 4. The results show that effects of sleep deprivation as well as trends within the sleep periods are readily apparent from spectral analysis, but are inadequately reflected by conventional sleep scoring. When the power density values were integrated over the entire frequency range (0.75-25 Hz) for each non-REM-REM sleep cycle, an exponential decline from cycle 1 to cycle 3 was suggested. The present findings support the hypothesis that the EEG power density in the low frequency range is an indicator of a progressively declining process during sleep whose initial value is determined by the duration of prior waking.

Low resolution brain electromagnetic tomography (LORETA) functional imaging in acute, neuroleptic-naive, first-episode, productive schizophrenia.

Functional imaging of brain electrical activity was performed in nine acute, neuroleptic-naive, first-episode, productive patients with schizophrenia and 36 control subjects. Low-resolution electromagnetic tomography (LORETA, three-dimensional images of cortical current density) was computed from 19-channel electroencephalographic (EEG) activity obtained under resting conditions, separately for the different EEG frequencies. Three patterns of activity were evident in the patients: (1) an anterior, near-bilateral excess of delta frequency activity; (2) an anterior-inferior deficit of theta frequency activity coupled with an anterior-inferior left-sided deficit of alpha-1 and alpha-2 frequency activity; and (3) a posterior-superior right-sided excess of beta-1, beta-2 and beta-3 frequency activity. Patients showed deviations from normal brain activity as evidenced by LORETA along an anterior-left-to-posterior-right spatial axis. The high temporal resolution of EEG makes it possible to specify the deviations not only as excess or deficit, but also as inhibitory, normal and excitatory. The patients showed a dis-coordinated brain functional state consisting of inhibited prefrontal/frontal areas and simultaneously overexcited right parietal areas, while left anterior, left temporal and left central areas lacked normal routine activity. Since all information processing is brain-state dependent, this dis-coordinated state must result in inadequate treatment of (externally or internally generated) information.

Segmentation of brain electrical activity into microstates: model estimation and validation

A brain microstate is defined as a functional/physiological state of the brain during which specific neural computations are performed. It is characterized uniquely by a fixed spatial distribution of active neuronal generators with time varying intensity. Brain electrical activity is modeled as being composed of a time sequence of nonoverlapping microstates with variable duration. A precise mathematical formulation of the model for evoked potential recordings is presented, where the microstates are represented as normalized vectors constituted by scalp electric potentials due to the underlying generators. An algorithm is developed for estimating the microstates, based on a modified version of the classical k-means clustering method, in which cluster orientations are estimated, Consequently, each instantaneous multichannel evoked potential measurement is classified as belonging to some microstate, thus producing a natural segmentation of brain activity. Use is made of statistical image segmentation techniques for obtaining smooth continuous segments. Time varying intensities are estimated by projecting the measurements onto their corresponding microstates. A goodness of fit statistic for the model is presented. Finally, a method is introduced for estimating the number of microstates, based on nonparametric data-driven statistical resampling techniques.<<ETX>>

EEG alpha map series: brain micro-states by space-oriented adaptive segmentation☆

The spontaneous EEG, viewed as a series of momentary scalp field maps, shows stable map configurations (of periodically reversed polarity) for varying durations, and discontinuous changes of the configurations. For adaptive segmentation of map series into spatially stationary epochs, the maps at the times of maximal map relief are selected and spatially described by the two locations of maximal and minimal (extreme) potentials; a segment ends if over time an extreme leaves its pre-set spatial window. Over 6 subjects, the resting alpha EEG showed 210 msec mean segment duration; segments longer than 323 msec covered 50% of the total time; the most prominent segment class (1.5% of all classes) covered 20% of total time (prominence varied strongly over classes; not all possible classes occurred). Spectral power and phase of averages of adaptive and pre-determined segments demonstrated the adequacy of the strategy, and the homogeneity of adaptive segment classes by their reduced within-class variance. It is suggested that different segment classes manifest different brain functional states exerting different effects on information processing. The spatially stationary segments might be basic building blocks of brain information processing, possibly operationalizing consciousness time and offering a common phenomenology for spontaneous activity and event-related potentials. The functional significance of segments might be modes or steps of information processing or performance, tested, e.g., as reaction time.

Spatial analysis of evoked potentials in man—a review

Steps in brain information processing are reflected on the scalp as changes of the electric potential which is evoked by the stimulus. However, for a given recording point on the scalp, there is no absolute amplitude or phase information of the electric brain potential. This means that the shape of an evoked potential waveform which is recorded from a given scalp location crucially depends on the location of the chosen reference. Only unbiased results of evoked potential data evaluation can be hoped to elucidate or map successfully into information processing models established by other methods, e.g. behavior measurements. Conventional recordings vs a common reference contain only one of many possible sets of waveshapes. In order to avoid ambiguities or bias of results, the entire evoked potential data set firstly must be analysed over space, and reference-independent parameters must be extracted. For each time point, the spatial distribution of the potentials is viewed as field map. The parameter extraction in a direct approach at each time point includes, e.g. locations of field peaks and troughs, voltage and gradient between them, and global electrical field power; or, parameters via the first or second spatial derivative of the electric field. In the second step, changes of these reference-independent field measurements are analysed over time. At component latency which is defined by maximal, global field power or by voltage range, mapped field distributions can be compared using maximal/minimal field value locations or complete maps. Significantly different field configurations establish the activity of non-identical neural generators. Classification of the field configurations (examination of orbits of field extrema over time) leads to the segmentation of series of field maps (multichannel EP data) into short epochs of stationary spatial configurations (i.e. spatially characterized components) with equal consideration of all recording points, and without the amplitude criterion. The application of these principles to the following problems is discussed: comparison of evoked potentials between different analysis times, in particular pre-stimulus and post-stimulus electric brain states; zero baseline for measurement; reference electrode; identification of evoked components in time and space. Illustrations of these problems include functional differences of input-analysing sub-systems, and the topography of cognition- and speech-related electric brain activity.

Affective Judgments of Faces Modulate Early Activity (160 ms) within the Fusiform Gyri

Functional neuroimaging studies have implicated the fusiform gyri (FG) in structural encoding of faces, while event-related potential (ERP) and magnetoen- cephalography studies have shown that such encoding occurs approximately 170 ms poststimulus. Behavioral and functional neuroimaging studies suggest that pro- cesses involved in face recognition may be strongly modulated by socially relevant information conveyed by faces. To test the hypothesis that affective informa- tion indeed modulates early stages of face processing, ERPs were recorded to individually assessed liked, neutral, and disliked faces and checkerboard-reversal stimuli. At the N170 latency, the cortical three-dimen- sional distribution of current density was computed in stereotactic space using a tomographic source local- ization technique. Mean activity was extracted from the FG, defined by structure-probability maps, and a meta-cluster delineated by the coordinates of the voxel with the strongest face-sensitive response from five published functional magnetic resonance imaging studies. In the FG, 160 ms poststimulus, liked faces elicited stronger activation than disliked and neutral faces and checkerboard-reversal stimuli. Further, confirming recent results, affect-modulated brain elec- trical activity started very early in the human brain (112 ms). These findings suggest that affective fea- tures conveyed by faces modulate structural face en- coding. Behavioral results from an independent study revealed that the stimuli were not biased toward par- ticular facial expressions and confirmed that liked faces were rated as more attractive. Increased FG ac- tivation for liked faces may thus be interpreted as reflecting enhanced attention due to their saliency.

Rapid emotional face processing in the human right and left brain hemispheres : an ERP study

Imaging work has begun to elucidate the spatial organization of emotions; the temporal organization, however, remains unclear. Adaptive behavior relies on rapid monitoring of potentially salient cues (typically with high emotional value) in the environment. To clarify the timing and speed of emotional processing in the two human brain hemispheres, event-related potentials (ERPs) were recorded during hemifield presentation of face images. ERPs were separately computed for disliked and liked faces, as individually assessed by postrecording affective ratings. After stimulation of either hemisphere, personal affective judgements of face images significantly modulated ERP responses at early stages, 80-116 ms after right hemisphere and 104-160 ms after left hemisphere stimulation. This is the first electrophysiological evidence for valence-dependent, automatic, i.e. pre-attentive emotional processing in humans.

Millisecond by millisecond, year by year: normative EEG microstates and developmental stages.

Most studies of continuous EEG data have used frequency transformation, which allows the quantification of brain states that vary over seconds. For the analysis of shorter, transient EEG events, it is possible to identify and quantify brain electric microstates as subsecond time epochs with stable field topography. These microstates may correspond to basic building blocks of human information processing. Microstate analysis yields a compact and comprehensive repertoire of short lasting classes of brain topographic maps, which may be considered to reflect global functional states. Each microstate class is described by topography, mean duration, frequency of occurrence and percentage analysis time occupied. This paper presents normative microstate data for resting EEG obtained from a database of 496 subjects between the age of 6 and 80 years. The extracted microstate variables showed a lawful, complex evolution with age. The pattern of changes with age is compatible with the existence of developmental stages as claimed by developmental psychologists. The results are discussed in the framework of state dependent information processing and suggest the existence of biologically predetermined top-down processes that bias brain electric activity to functional states appropriate for age-specific learning and behavior.

Assessing interactions in the brain with exact low-resolution electromagnetic tomography

Scalp electric potentials (electroencephalogram; EEG) are contingent to the impressed current density unleashed by cortical pyramidal neurons undergoing post-synaptic processes. EEG neuroimaging consists of estimating the cortical current density from scalp recordings. We report a solution to this inverse problem that attains exact localization: exact low-resolution brain electromagnetic tomography (eLORETA). This non-invasive method yields high time-resolution intracranial signals that can be used for assessing functional dynamic connectivity in the brain, quantified by coherence and phase synchronization. However, these measures are non-physiologically high because of volume conduction and low spatial resolution. We present a new method to solve this problem by decomposing them into instantaneous and lagged components, with the lagged part having almost pure physiological origin.