Sidse M. Arnfred

Parallel Factor Analysis as an exploratory tool for wavelet transformed event-related EEG

In the decomposition of multi-channel EEG signals, principal component analysis (PCA) and independent component analysis (ICA) have widely been used. However, as both methods are based on handling two-way data, i.e. two-dimensional matrices, multi-way methods might improve the interpretation of frequency transformed multi-channel EEG of channel x frequency x time data. The multi-way decomposition method Parallel Factor (PARAFAC), also named Canonical Decomposition (CANDECOMP), was recently used to decompose the wavelet transformed ongoing EEG of channel x frequency x time (Miwakeichi, F., Martinez-Montes, E., Valdes-Sosa, P.A., Nishiyama, N., Mizuhara, H., Yamaguchi, Y., 2004. Decomposing EEG data into space-time-frequency components using parallel factor analysis. Neuroimage 22, 1035-1045). In this article, PARAFAC is used for the first time to decompose wavelet transformed event-related EEG given by the inter-trial phase coherence (ITPC) encompassing ANOVA analysis of differences between conditions and 5-way analysis of channel x frequency x time x subject x condition. A flow chart is presented on how to perform data exploration using the PARAFAC decomposition on multi-way arrays. This includes (A) channel x frequency x time 3-way arrays of F test values from a repeated measures analysis of variance (ANOVA) between two stimulus conditions; (B) subject-specific 3-way analyses; and (C) an overall 5-way analysis of channel x frequency x time x subject x condition. The PARAFAC decompositions were able to extract the expected features of a previously reported ERP paradigm: namely, a quantitative difference of coherent occipital gamma activity between conditions of a visual paradigm. Furthermore, the method revealed a qualitative difference which has not previously been reported. The PARAFAC decomposition of the 3-way array of ANOVA F test values clearly showed the difference of regions of interest across modalities, while the 5-way analysis enabled visualization of both quantitative and qualitative differences. Consequently, PARAFAC is a promising data exploratory tool in the analysis of the wavelets transformed event-related EEG.

Algorithms for sparse nonnegative tucker decompositions

There is a increasing interest in analysis of large-scale multiway data. The concept of multiway data refers to arrays of data with more than two dimensions, that is, taking the form of tensors. To analyze such data, decomposition techniques are widely used. The two most common decompositions for tensors are the Tucker model and the more restricted PARAFAC model. Both models can be viewed as generalizations of the regular factor analysis to data of more than two modalities. Nonnegative matrix factorization (NMF), in conjunction with sparse coding, has recently been given much attention due to its part-based and easy interpretable representation. While NMF has been extended to the PARAFAC model, no such attempt has been done to extend NMF to the Tucker model. However, if the tensor data analyzed are nonnegative, it may well be relevant to consider purely additive (i.e., nonnegative) Tucker decompositions). To reduce ambiguities of this type of decomposition, we develop updates that can impose sparseness in any combination of modalities, hence, proposed algorithms for sparse nonnegative Tucker decompositions (SN-TUCKER). We demonstrate how the proposed algorithms are superior to existing algorithms for Tucker decompositions when the data and interactions can be considered nonnegative. We further illustrate how sparse coding can help identify what model (PARAFAC or Tucker) is more appropriate for the data as well as to select the number of components by turning off excess components. The algorithms for SN-TUCKER can be downloaded from Mrup (2007).

ERPWAVELAB a toolbox for multi-channel analysis of time-frequency transformed event related potentials.

The open source toolbox ERPWAVELAB is developed for multi-channel time-frequency analysis of event related activity of EEG and MEG data. The toolbox provides tools for data analysis and visualization of the most commonly used measures of time-frequency transformed event related data as well as data decomposition through non-negative matrix and multi-way (tensor) factorization. The decompositions provided can accommodate additional dimensions like subjects, conditions or repeats and as such they are perfected for group analysis. Furthermore, the toolbox enables tracking of phase locked activity from one channel-time-frequency instance to another as well as tools for artifact rejection in the time-frequency domain. Several other features are highlighted. ERPWAVELAB can freely be downloaded from www.erpwavelab.org, requires EEGLAB [Delorme A, Makeig S. EEGLAB: an open source toolbox for analysis of single-trial EEG dynamics including independent component analysis. J Neurosci Meth 2004;134:9-21] and runs under MATLAB (The Mathworks, Inc.).

Shift-invariant multilinear decomposition of neuroimaging data

We present an algorithm for multilinear decomposition that allows for arbitrary shifts along one modality. The method is applied to neural activity arranged in the three modalities space, time, and trial. Thus, the algorithm models neural activity as a linear superposition of components with a fixed time course that may vary across either trials or space in its overall intensity and latency. Its utility is demonstrated on simulated data as well as actual EEG, and fMRI data. We show how shift-invariant multilinear decompositions of multiway data can successfully cope with variable latencies in data derived from neural activity--a problem that has caused degenerate solutions especially in modeling neuroimaging data with instantaneous multilinear decompositions. Our algorithm is available for download at www.erpwavelab.org.

The amplitude and phase precision of 40 Hz auditory steady-state response depend on the level of arousal.

The aim of this study was to investigate, in healthy subjects, the modulation of amplitude and phase precision of the auditory steady-state response (ASSR) to 40xa0Hz stimulation in two resting conditions varying in the level of arousal. Previously, ASSR measures have shown to be affected by the level of arousal, but the findings are somewhat controversial. Generally, ASSR is diminished in sleep but it may be increased in drowsiness. Besides, ASSR reduction has been observed in schizophrenia. However, schizophrenic patients are known to have a disturbance of arousal level, what makes it pertinent to know the effects of fluctuations in arousal on passive response to gamma-range stimulation. In nine healthy volunteers trains of 40xa0Hz click stimuli were applied during two conditions: in the “high arousal” condition subjects were sitting upright silently reading a book of interest; in the “low arousal” condition subjects were sitting in a reclined position with eyes closed and the lights turned off. The 64-channel EEG data was wavelet transformed and the amplitude and phase precision of the wavelet transformed evoked potential were decomposed by the recently proposed multi-subject non-negative multi-way factorization (NMWF) (Morup et al. in J Neurosci Methods 161:361–368, 2007). The estimates of these measures were subjected to statistical analysis. The amplitude and phase precision of the ASSR were significantly larger during the low arousal state compared to the high arousal condition. The modulation of ASSR amplitude and phase precision by differences in the arousal level during recording warrants caution when investigating oscillatory brain activity and interpreting the findings of reduced ASSR in schizophrenia. It also emphasizes the necessity of standardized recording procedures and monitoring the level of arousal during ASSR testing.

Exploration of somatosensory P50 gating in schizophrenia spectrum patients: reduced P50 amplitude correlates to social anhedonia

Originally, the hypothesis of a sensory gating defect in schizophrenia evolved from studies of somatosensory evoked potentials (SEP), although the idea has primarily been pursued in the auditory modality. Gating is the relative attenuation of amplitude following the second stimulus in a stimulus pair. Recently, SEP P50 gating was seen when recording the SEP P50 in a paradigm similar to the one used for auditory P50 gating. Hypothetically, abnormality of somatosensory information processing could be related to anhedonia, which is considered a core feature of schizophrenia. Twelve unmedicated, male, schizophrenia spectrum patients (seven schizophrenic and five schizotypal personality disorder patients) and 14 age-matched healthy men participated in recordings of pair-wise presented auditory and median nerve stimuli. The patients had smaller amplitudes of the SEP P50 at the first stimulus, but no gating defect. The reduced amplitude was particularly evident in subjects with high scores on the Revised Social Anhedonia Scale. Early somatosensory information processing seems abnormal in schizophrenia spectrum patients. This could be in agreement with the theory of loss of the benefit of regularity in schizophrenia, while the results are in-conclusive regarding sensory gating theory.

Distraction task rather than focal attention modulates gamma activity associated with auditory steady-state responses (ASSRs)

OBJECTIVEnTo explore the modulation of auditory steady-state response (ASSR) by experimental tasks, differing in attentional focus and arousal level.nnnMETHODSn20 Hz and 40 Hz click trains were used to elicit ASSRs. Experiment 1 consisted of two runs of closed eyes and two runs of open eyes. Experiment 2 consisted of six tasks: counting 20 Hz and 40 Hz stimuli, sitting with closed and open eyes, reading an article, and performing a search task. Phase locking factor, evoked amplitude and total intensity were decomposed by non-negative multi-way factorization.nnnRESULTSnThe total intensity of 40 Hz ASSR was enhanced during closed eyes condition in comparison to the open eyes condition. The evoked amplitude and phase locking factor of 40 Hz ASSR were attenuated during distraction, while there were no differences between attention to stimulation and no task. 20 Hz ASSR and 20 Hz ASSR-related 40 Hz activity were not modulated by the tasks.nnnCONCLUSIONSnThe phase-locked measures of 40 Hz ASSR are attenuated when attention to the stimulation is low, i.e. the subject is effortfully focused on the competitive distraction task performance. Lower arousal level increases the total intensity of 40 Hz ASSRs.nnnSIGNIFICANCEnImprovements of the practical use of ASSRs are suggested: a careful monitoring for arousal fluctuations during ASSR recordings should be performed; when ASSRs are applied to investigate the ability to generate high frequency cortical activity a distraction task is not favorable.

Attenuation of beta and gamma oscillations in schizophrenia spectrum patients following hand posture perturbation

Several electroencephalographic (EEG) studies in schizophrenia report that the patients have reduced evoked gamma activity following visual and auditory stimulation. Somatosensory gamma activity has not previously been examined. It has been suggested that a dysfunction basic to schizophrenia spectrum traits would involve proprioceptive information processing and this has recently been supported by the finding of diminished latency of early proprioceptive evoked potentials in a sample of chronic schizophrenia patients. The proprioceptive stimulus used previously, and presently, consisted of an abrupt increase of weight on a hand-held load. Eighteen first-time admitted schizophrenia spectrum patients and 18 healthy matched comparison subjects were included. Proprioceptive evoked potentials were recorded as 64-channels EEG for 120 trials in two runs differing in sequence. Contra-lateral evoked beta (latency 90 ms, frequency 21 Hz) and gamma (latency 70 ms, frequency 32 Hz) oscillations were attenuated in the patient group. The healthy comparison subjects had increased gamma amplitude in the left hemisphere in the regular sequence, a phenomenon not seen in the patients. The deviant findings were unexpectedly more circumscribed in the schizophrenia than in the schizotypal personality disorder (SPD) patients. Future studies should include several concurrent psychophysiological measures.

Two discrete components of the 20 Hz steady-state response are distinguished through the modulation of activation level

OBJECTIVEnTo investigate the modulation of amplitude and phase precision of the auditory steady-state response (SSR) to 20 Hz stimulation in two conditions varying in the level of activation.nnnMETHODSnClick stimuli (20 Hz) were applied while subjects were sitting upright silently reading a book of interest (high activation level) and while subjects were sitting in a reclined position with eyes closed and the lights turned off (low activation level). Sixty-one channel EEG data was wavelet transformed, the amplitude and phase precision measures extracted and decomposed by the multi-subject non-negative multi-way factorization (NMWF).nnnRESULTSnThe NMWF decomposition of amplitude and phase precision measures resulted in the observation of two distinct components: a component at the frequency of stimulation--20 Hz SSR and a component emerging at 40 Hz--20 Hz SSR-related 40 Hz activity. Modulation by the activation level was observed only for 20 Hz SSR-related 40 Hz activity as increased amplitude and phase precision during low activation level. No such effects were observed for 20 Hz SSR.nnnCONCLUSIONnThe discrete components of the 20 Hz SSR are distinguished through modulation of activation level, 20 Hz SSR-related 40 Hz being higher in low activation state.nnnSIGNIFICANCEnThe biological modulation of 20 Hz SSR-related 40 Hz activity by the level of activation points to a physiological nature of this activity beyond a mere periodic effect in relation to the 20 Hz activity.

Regularity increases middle latency evoked and late induced beta brain response following proprioceptive stimulation.

Focal attention increases the middle-latency amplitude of somatosensory evoked potentials. Previously this effect has been suggested to be due to increased readiness in somatosensory cortex. Presently, we examine whether regularity of stimulus occurrence increases the proprioceptive evoked response as an indication of increased readiness. This is achieved through detailed analysis of both evoked and induced responses in the time-frequency domain. Electroencephalography in a 64 channels montage was recorded in fourteen healthy subjects. Two paradigms were explored: A Regular alternation between hand of presentation and a Random sequence of hand of presentation. The ERPWAVELAB toolbox was used for decomposition of the wavelet transformed data (7 to 47 Hz, -300 to +1500 ms) yielding the evoked amplitude (AvWT) and inter-trial phase coherence as well as the increase of non-time-locked activity (Induced). After initial exploration of the AvWT and Induced collapsed files of all subjects using two-way factor analyses (Non-Negative Matrix Factorization), further data decomposition was performed in restricted windows of interest (WOI). Main effects of side of stimulation, onset or offset, regularity and habituation on the evoked and induced activity are described for each WOI. The Regular paradigm evoked more activity than Random in the fast beta range (18-28 Hz) (mean: normalized amplitude 0.38 at 90 ms and 20.9 Hz) including increased phase precision. The findings confirm the possibility of modulation of middle-latency activity by regularity and suggest that this is due to facilitation of activity in secondary somatosensory cortices. Future studies need to examine whether the increased amplitude is associated with increased perceptual acuity.