Interaction between steady-state visually evoked potentials at nearby flicker frequencies

Alpha and SSVEP power outperform gamma power in capturing attentional modulation in human EEG

Attention typically reduces power in the alpha (8-12 Hz) band and increases power in gamma (>30 Hz) band in brain signals, as reported in macaque local field potential (LFP) and human electro/magneto-encephalogram (EEG/MEG) studies. In addition, EEG studies often use flickering stimuli that produce a specific measure called steady-state-visually-evoked-potential (SSVEP), whose power also increases with attention. However, effectiveness of these neural measures in capturing attentional modulation is unknown since stimuli and task paradigms vary widely across studies. In a recent macaque study, attentional modulation was more salient in the gamma band of the LFP, compared to alpha or SSVEP. To compare this with human EEG, we designed an orientation change detection task where we presented both static and counterphasing stimuli of matched difficulty levels to 26 subjects and compared attentional modulation of various measures under similar conditions. We report two main results. First, attentional modulation was comparable for SSVEP and alpha. Second, non-foveal stimuli produced weak gamma despite various stimulus optimizations and showed negligible attentional modulation although full-screen gratings showed robust gamma activity. Our results are useful for brain-machine-interfacing studies where suitable features are used for decoding attention, and also provide clues about spatial scales of neural mechanisms underlying attention.

EEG-FMCNN: A fusion multi-branch 1D convolutional neural network for EEG-based motor imagery classification

Motor imagery (MI) electroencephalogram (EEG) signal is recognized as a promising paradigm for brain-computer interface (BCI) systems and has been extensively employed in various BCI applications, including assisting disabled individuals, controlling devices and environments, and enhancing human capabilities. The high-performance decoding capability of MI-EEG signals is a key issue that impacts the development of the industry. However, decoding MI-EEG signals is challenging due to the low signal-to-noise ratio and inter-subject variability. In response to the aforementioned core problems, this paper proposes a novel end-to-end network, a fusion multi-branch 1D convolutional neural network (EEG-FMCNN), to decode MI-EEG signals without pre-processing. The utilization of multi-branch 1D convolution not only exhibits a certain level of noise tolerance but also addresses the issue of inter-subject variability to some extent. This is attributed to the ability of multi-branch architectures to capture information from different frequency bands, enabling the establishment of optimal convolutional scales and depths. Furthermore, we incorporate 1D squeeze-and-excitation (SE) blocks and shortcut connections at appropriate locations to further enhance the generalization and robustness of the network. In the BCI Competition IV-2a dataset, our proposed model has obtained good experimental results, achieving accuracies of 78.82% and 68.41% for subject-dependent and subject-independent modes, respectively. In addition, extensive ablative experiments and fine-tuning experiments were conducted, resulting in a notable 7% improvement in the average performance of the network, which holds significant implications for the generalization and application of the network.

GDNet-EEG: An attention-aware deep neural network based on group depth-wise convolution for SSVEP stimulation frequency recognition

Background

Steady state visually evoked potentials (SSVEPs) based early glaucoma diagnosis requires effective data processing (e.g., deep learning) to provide accurate stimulation frequency recognition. Thus, we propose a group depth-wise convolutional neural network (GDNet-EEG), a novel electroencephalography (EEG)-oriented deep learning model tailored to learn regional characteristics and network characteristics of EEG-based brain activity to perform SSVEPs-based stimulation frequency recognition.

Method

Group depth-wise convolution is proposed to extract temporal and spectral features from the EEG signal of each brain region and represent regional characteristics as diverse as possible. Furthermore, EEG attention consisting of EEG channel-wise attention and specialized network-wise attention is designed to identify essential brain regions and form significant feature maps as specialized brain functional networks. Two publicly SSVEPs datasets (large-scale benchmark and BETA dataset) and their combined dataset are utilized to validate the classification performance of our model.

Results

Based on the input sample with a signal length of 1 s, the GDNet-EEG model achieves the average classification accuracies of 84.11, 85.93, and 93.35% on the benchmark, BETA, and combination datasets, respectively. Compared with the average classification accuracies achieved by comparison baselines, the average classification accuracies of the GDNet-EEG trained on a combination dataset increased from 1.96 to 18.2%.

Conclusion

Our approach can be potentially suitable for providing accurate SSVEP stimulation frequency recognition and being used in early glaucoma diagnosis.

Local Interactions between Steady-State Visually Evoked Potentials at Nearby Flickering Frequencies

Steady-state visually evoked potentials (SSVEPs) are widely used to index top-down cognitive processing in human electroencephalogram (EEG) studies. Typically, two stimuli flickering at different temporal frequencies (TFs) are presented, each producing a distinct response in the EEG at its flicker frequency. However, how SSVEP responses in EEGs are modulated in the presence of a competing flickering stimulus just because of sensory interactions is not well understood. We have previously shown in local field potentials (LFPs) recorded from awake monkeys that when two overlapping full-screen gratings are counterphased at different TFs, there is an asymmetric SSVEP response suppression, with greater suppression from lower TFs, which further depends on the relative orientations of the gratings (stronger suppression and asymmetry for parallel compared with orthogonal gratings). Here, we first confirmed these effects in both male and female human EEG recordings. Then, we mapped the response suppression of one stimulus (target) by a competing stimulus (mask) over a much wider range than the previous study. Surprisingly, we found that the suppression was not stronger at low frequencies in general, but systematically varied depending on the target TF, indicating local interactions between the two competing stimuli. These results were confirmed in both human EEG and monkey LFP and electrocorticogram (ECoG) data. Our results show that sensory interactions between multiple SSVEPs are more complex than shown previously and are influenced by both local and global factors, underscoring the need to cautiously interpret the results of studies involving SSVEP paradigms.SIGNIFICANCE STATEMENT Steady-state visually evoked potentials (SSVEPs) are extensively used in human cognitive studies and brain-computer interfacing applications where multiple stimuli flickering at distinct frequencies are concurrently presented in the visual field. We recently characterized interactions between competing flickering stimuli in animal recordings and found that stimuli flickering slowly produce larger suppression. Here, we confirmed these in human EEGs, and further characterized the interactions by using a much wider range of target and competing (mask) frequencies in both human EEGs and invasive animal recordings. These revealed a new "local" component, whereby the suppression increased when competing stimuli flickered at nearby frequencies. Our results highlight the complexity of sensory interactions among multiple SSVEPs and underscore the need to cautiously interpret studies involving SSVEP paradigms.

A 1D CNN for high accuracy classification and transfer learning in motor imagery EEG-based brain-computer interface

OBJECTIVE

Brain-computer interface (BCI) aims to establish communication paths between the brain processes and external devices. Different methods have been used to extract human intentions from electroencephalography (EEG) recordings. Those based on motor imagery (MI) seem to have a great potential for future applications. These approaches rely on the extraction of EEG distinctive patterns during imagined movements. Techniques able to extract patterns from raw signals represent an important target for BCI as they do not need labor-intensive data pre-processing.

APPROACH

We propose a new approach based on a 10-layer one-dimensional convolution neural network (1D-CNN) to classify five brain states (four MI classes plus a 'baseline' class) using a data augmentation algorithm and a limited number of EEG channels. In addition, we present a transfer learning method used to extract critical features from the EEG group dataset and then to customize the model to the single individual by training its outer layers with only 12-minute individual-related data.

MAIN RESULTS

The model tested with the 'EEG Motor Movement/Imagery Dataset' outperforms the current state-of-the-art models by achieving a 99.38% accuracy at the group level. In addition, the transfer learning approach we present achieves an average accuracy of 99.46%.

SIGNIFICANCE

The proposed methods could foster future BCI applications relying on few-channel portable recording devices and individual-based training.

Solving the SSVEP Paradigm Using the Nonlinear Canonical Correlation Analysis Approach

This paper presents the implementation of nonlinear canonical correlation analysis (NLCCA) approach to detect steady-state visual evoked potentials (SSVEP) quickly. The need for the fast recognition of proper stimulus to help end an SSVEP task in a BCI system is justified due to the flickering external stimulus exposure that causes users to start to feel fatigued. Measuring the accuracy and exposure time can be carried out through the information transfer rate-ITR, which is defined as a relationship between the precision, the number of stimuli, and the required time to obtain a result. NLCCA performance was evaluated by comparing it with two other approaches-the well-known canonical correlation analysis (CCA) and the least absolute reduction and selection operator (LASSO), both commonly used to solve the SSVEP paradigm. First, the best average ITR value was found from a dataset comprising ten healthy users with an average age of 28, where an exposure time of one second was obtained. In addition, the time sliding window responses were observed immediately after and around 200 ms after the flickering exposure to obtain the phase effects through the coefficient of variation (CV), where NLCCA obtained the lowest value. Finally, in order to obtain statistical significance to demonstrate that all approaches differ, the accuracy and ITR from the time sliding window responses was compared using a statistical analysis of variance per approach to identify differences between them using Tukey's test.

Impact of Visual Design Elements and Principles in Human Electroencephalogram Brain Activity Assessed with Spectral Methods and Convolutional Neural Networks

The visual design elements and principles (VDEPs) can trigger behavioural changes and emotions in the viewer, but their effects on brain activity are not clearly understood. In this paper, we explore the relationships between brain activity and colour (cold/warm), light (dark/bright), movement (fast/slow), and balance (symmetrical/asymmetrical) VDEPs. We used the public DEAP dataset with the electroencephalogram signals of 32 participants recorded while watching music videos. The characteristic VDEPs for each second of the videos were manually tagged for by a team of two visual communication experts. Results show that variations in the light/value, rhythm/movement, and balance in the music video sequences produce a statistically significant effect over the mean absolute power of the Delta, Theta, Alpha, Beta, and Gamma EEG bands (p < 0.05). Furthermore, we trained a Convolutional Neural Network that successfully predicts the VDEP of a video fragment solely by the EEG signal of the viewer with an accuracy ranging from 0.7447 for Colour VDEP to 0.9685 for Movement VDEP. Our work shows evidence that VDEPs affect brain activity in a variety of distinguishable ways and that a deep learning classifier can infer visual VDEP properties of the videos from EEG activity.

Stimulus-induced gamma rhythms are weaker in human elderly with mild cognitive impairment and Alzheimer's disease

Alzheimer's disease (AD) in elderly adds substantially to socioeconomic burden necessitating early diagnosis. While recent studies in rodent models of AD have suggested diagnostic and therapeutic value for gamma rhythms in brain, the same has not been rigorously tested in humans. In this case-control study, we recruited a large population (N = 244; 106 females) of elderly (>49 years) subjects from the community, who viewed large gratings that induced strong gamma oscillations in their electroencephalogram (EEG). These subjects were classified as healthy (N = 227), mild cognitively impaired (MCI; N = 12), or AD (N = 5) based on clinical history and Clinical Dementia Rating scores. Surprisingly, stimulus-induced gamma rhythms, but not alpha or steady-state visually evoked responses, were significantly lower in MCI/AD subjects compared to their age- and gender-matched controls. This reduction was not due to differences in eye movements or baseline power. Our results suggest that gamma could be used as a potential screening tool for MCI/AD in humans.

Effect of Cross-Orientation Normalization on Different Neural Measures in Macaque Primary Visual Cortex

Divisive normalization is a canonical mechanism that can explain a variety of sensory phenomena. While normalization models have been used to explain spiking activity in response to different stimulus/behavioral conditions in multiple brain areas, it is unclear whether similar models can also explain modulation in population-level neural measures such as power at various frequencies in local field potentials (LFPs) or steady-state visually evoked potential (SSVEP) that is produced by flickering stimuli and popular in electroencephalogram studies. To address this, we manipulated normalization strength by presenting static as well as flickering orthogonal superimposed gratings (plaids) at varying contrasts to 2 female monkeys while recording multiunit activity (MUA) and LFP from the primary visual cortex and quantified the modulation in MUA, gamma (32-80 Hz), high-gamma (104-248 Hz) power, as well as SSVEP. Even under similar stimulus conditions, normalization strength was different for the 4 measures and increased as: spikes, high-gamma, SSVEP, and gamma. However, these results could be explained using a normalization model that was modified for population responses, by varying the tuned normalization parameter and semisaturation constant. Our results show that different neural measures can reflect the effect of stimulus normalization in different ways, which can be modeled by a simple normalization model.

The SSVEP tracks attention, not consciousness, during perceptual filling-in

Research on the neural basis of conscious perception has almost exclusively shown that becoming aware of a stimulus leads to increased neural responses. By designing a novel form of perceptual filling-in (PFI) overlaid with a dynamic texture display, we frequency-tagged multiple disappearing targets as well as their surroundings. We show that in a PFI paradigm, the disappearance of a stimulus and subjective invisibility is associated with increases in neural activity, as measured with steady-state visually evoked potentials (SSVEPs), in electroencephalography (EEG). We also find that this increase correlates with alpha-band activity, a well-established neural measure of attention. These findings cast doubt on the direct relationship previously reported between the strength of neural activity and conscious perception, at least when measured with current tools, such as the SSVEP. Instead, we conclude that SSVEP strength more closely measures changes in attention.

Gamma oscillations weaken with age in healthy elderly in human EEG

Gamma rhythms (~20-70 ​Hz) are abnormal in mental disorders such as autism and schizophrenia in humans, and Alzheimer's disease (AD) models in rodents. However, the effect of normal aging on these oscillations is unknown, especially for elderly subjects in whom AD is most prevalent. In a first large-scale (236 subjects; 104 females) electroencephalogram (EEG) study on gamma oscillations in elderly subjects (aged 50-88 years), we presented full-screen visual Cartesian gratings that induced two distinct gamma oscillations (slow: 20-34 ​Hz and fast: 36-66 ​Hz). Power decreased with age for gamma, but not alpha (8-12 ​Hz). Reduction was more salient for fast gamma than slow. Center frequency also decreased with age for both gamma rhythms. The results were independent of microsaccades, pupillary reactivity to stimulus, and variations in power spectral density with age. Steady-state visual evoked potentials (SSVEPs) at 32 ​Hz also reduced with age. These results are crucial for developing gamma/SSVEP-based biomarkers of cognitive decline in elderly.