Robert Isenhart

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.

Electrophysiological analysis of the registration, storage and retrieval of information in delayed matching from samples.

Brain processes of registration, storage in working memory and retrieval of different kinds of information were studied by analysis of EEG and ERP activity recorded during two delayed matches from sample tasks: (1) matching the digits in two series of six numbers, and (2) matching the sums of the same two series of six numbers. each trial was composed of six intervals continuing six equally spaced visual stimuli: (1) control--observing a series of six fixation points, P1, on a computer monitor; (2) priming--viewing a series, S1, of six numbers; (3) delay--observing a second series of six fixation points, P2; (4) matching--viewing a second series, S2, of six numbers; (5) response selection--selecting the left button to press if S1 contained all the items in S2 or the right button if any item appeared only in S2, while observing six fixation points; (6) feedback--six color coded fixation points indicate correct or error. Each interval was 4 s in duration and 20 trials were presented in each task. During each interval the visual field flickered at a tracer frequency of 1.5/s, whether numbers or fixation points were on the monitor screen. Very narrow band power spectra (VNB), ERPs elicited by presentation of S1 or S2 information items, and non-contingent probes (NCP) elicited by presentation of fixation points were used to trace the processing of information by neural populations activated by the visual stimulation. Global field power maxima identified latencies at which functional landscapes were analyzed. VNB, ERP, NCP and landscape differences were found between digits and sums. However, though these differences were highly significant within each subject (p < 0.001), no consistency was found across individuals for the electrophysiological changes during the tasks. This suggests that utilization of brain resources in cognition varies greatly with individual cognitive styles and strategies.

Localizing epileptogenic regions in partial epilepsy using three-dimensional statistical parametric maps of background EEG source spectra.

This preliminary study sought to localize epileptogenic regions in patients with partial epilepsy by analysis of interictal EEG activity utilizing variable resolution electromagnetic tomography (VARETA), a three-dimensional quantitative electroencephalographic (QEEG) frequency-domain distributed source modeling technique. The very narrow band (VNB) spectra spanned the frequency range 0.39 Hz to 19.1 Hz, in 0.39 Hz steps. These VNB spectra were compared to normative data and transformed to provide Z-scores for every scalp derivation, and the spatial distributions of the probable EEG generators of the most abnormal values were displayed on slices from a probabilistic MRI atlas. Each voxel was color-coded to represent the significance of the deviation relative to age appropriate normative values. We compared the resulting three-dimensional images to the localization of epileptogenic regions based on invasive intracranial EEG recordings of seizure onsets. The VARETA image indicated abnormal interictal spectral power values in regions of seizure onset identified by invasive monitoring, mainly in delta and theta range (1.5 to 8.0 Hz). The VARETA localization of the most abnormal voxel was congruent with the epileptogenic regions identified by intracranial recordings with regard to hemisphere in all 6 cases, and with regard to lobe in 5 cases. In contrast, abnormal findings with routine EEG agreed with invasive monitoring with regard to hemisphere in 3 cases and with regard to lobe in 2 cases. These results suggest that analysis of background interictal EEG utilizing distributed source models should be investigated further in clinical epilepsy.

Quantitative EEG Tomography of Early Childhood Malnutrition

The goal of this study is to identify the quantitative electroencephalographic (qEEG) signature of early childhood malnutrition [protein-energy malnutrition (PEM)]. To this end, archival digital EEG recordings of 108 participants in the Barbados Nutrition Study (BNS) were recovered and cleaned of artifacts (46 children who suffered an episode of PEM limited to the first year of life) and 62 healthy controls). The participants of the still ongoing BNS were initially enrolled in 1973, and EEGs for both groups were recorded in 1977–1978 (at 5–11 years). Scalp and source EEG Z-spectra (to correct for age effects) were obtained by comparison with the normative Cuban Human Brain Mapping database. Differences between both groups in the z spectra (for all electrode locations and frequency bins) were assessed by t-tests with thresholds corrected for multiple comparisons by permutation tests. Four clusters of differences were found: (a) increased theta activity (3.91–5.86 Hz) in electrodes T4, O2, Pz and in the sources of the supplementary motor area (SMA); b) decreased alpha1 (8.59–8.98 Hz) in Fronto-central electrodes and sources of widespread bilateral prefrontal are; (c) increased alpha2 (11.33–12.50 Hz) in Temporo-parietal electrodes as well as in sources in Central-parietal areas of the right hemisphere; and (d) increased beta (13.67–18.36 Hz), in T4, T5 and P4 electrodes and decreased in the sources of bilateral occipital-temporal areas. Multivariate Item Response Theory of EEGs scored visually by experts revealed a neurophysiological latent variable which indicated excessive paroxysmal and focal abnormality activity in the PEM group. A robust biomarker construction procedure based on elastic-net regressions and 1000-cross-validations was used to: (i) select stable variables and (ii) calculate the area under ROC curves (AUC). Thus, qEEG differentiate between the two nutrition groups (PEM vs Control) performing as well as visual inspection of the EEG scored by experts (AUC = 0.83). Since PEM is a global public health problem with lifelong neurodevelopmental consequences, our finding of consistent differences between PEM and controls, both in qualitative and quantitative EEG analysis, suggest that this technology may be a source of scalable and affordable biomarkers for assessing the long-term brain impact of early PEM.

F168. An EEG fingerprint of early protein-energy malnutrition

Introduction Early childhood Protein Energy Malnutrition (PEM) is an increasing worldwide phenomenon with lifelong neurodevelopmental consequences. There is thus a need for inexpensive imaging technologies to objectively identify and follow up the neural impact of malnutrition—Electroencephalography being an obvious choice. But EEG studies of PEM are scarce, performed on subjects with multiple stressors, only in the acute phase. A unique opportunity to improve these enquiries is the still ongoing Barbados Nutrition Study (BNS) which enrolled (1967–72) children with PEM during their first year of life. Under the direction of Frank Ramsey and E. Roy John, 248 digital EEG recordings were obtained (children 5–11 years) at the time that the Brain Research Lab and the Cuban Neuroscience Centre were developing quantitative EEG (qEEG; John et al., 1977). Recently, a large subsample of these digital EEGs was recovered. A unique opportunity to identify a qEEG fingerprint of early PEM has thus arisen, and which we here report. Methods The final sample comprised 46 PEM and 62 control recordings (1 min resting state, eyes-closed,19 electrodes 10/20 system, sampling 100 Hz). Qualitative EEG was evaluated using a Likert-type scale. Multivariate Item Response Theory identified a neurophysiological state (NS) as a single latent variable explaining 0.88 of sample variance. qEEG evaluation at the electrodes (topography) consisted in calculating the log-power spectrum both at the scalp electrodes and sources and computing the z transform with regard to the Cuban normative database. Quantitative tomographic EEG (qEEGt) was carried out with CNEURO’s VARETA source analysis procedure based upon an MNI probabilistic template—necessary since MRIs where not available at that time. Multivariate permutation tests ( N  = 1000) were applied to t -tests in order to assess differences between groups. Results Qualitative analysis revealed highly significant changes in the latent variable (NS) with the PEM group showing excessive slow-wave, paroxysmal and focal abnormality activity, with a statistically significant effect for groups ( p 15.2 Hz. qEEGT analysis: the PEM group, showed a significant increment in source power at lower frequencies ( Conclusion The consistent differences in qEEG and qEEGt values between PEM and controls suggest they may be affordable biomarkers for the long-term actual brain impact of early childhood PEM. Excess slow-waves activity and decreased alpha activity in PEM children, may be a qEEG fingerprint of early PEM predicting which is correlated with many types of neuropathology, learning and performance difficulties.