Harmonized-Multinational qEEG norms (HarMNqEEG).

This paper extends frequency domain quantitative electroencephalography (qEEG) methods pursuing higher sensitivity to detect Brain Developmental Disorders. Prior qEEG work lacked integration of cross-spectral information omitting important functional connectivity descriptors. Lack of geographical diversity precluded accounting for site-specific variance, increasing qEEG nuisance variance. We ameliorate these weaknesses. (i) Create lifespan Riemannian multinational qEEG norms for cross-spectral tensors. These norms result from the HarMNqEEG project fostered by the Global Brain Consortium. We calculate the norms with data from 9 countries, 12 devices, and 14 studies, including 1564 subjects. Instead of raw data, only anonymized metadata and EEG cross-spectral tensors were shared. After visual and automatic quality control, developmental equations for the mean and standard deviation of qEEG traditional and Riemannian DPs were calculated using additive mixed-effects models. We demonstrate qEEG "batch effects" and provide methods to calculate harmonized z-scores. (ii) We also show that harmonized Riemannian norms produce z-scores with increased diagnostic accuracy predicting brain dysfunction produced by malnutrition in the first year of life and detecting COVID induced brain dysfunction. (iii) We offer open code and data to calculate different individual z-scores from the HarMNqEEG dataset. These results contribute to developing bias-free, low-cost neuroimaging technologies applicable in various health settings.

The Statistics of EEG Unipolar References: Derivations and Properties.

In this brief communication, which complements the EEG reference review (Yao et al. in Brain Topogr, 2019), we provide the mathematical derivations that show: (1) any EEG reference admits the general form of a linear transformation of the ideal multichannel EEG potentials with reference to infinity; (2) the average reference (AR), the reference electrode standardization technique (REST), and its regularized version (rREST) are solving the linear inverse problems that can be derived from both the maximum likelihood estimate (MLE) and the Bayesian theory; however, REST is based on more informative prior/constraint of volume conduction than that of AR; (3) we show for the first time that REST is also a unipolar reference (UR), allowing us to define a general family of URs with unified notations; (4) some notable properties of URs are 'no memory', 'rank deficient by 1', and 'orthogonal projector centering'; (5) we also point out here, for the first time, that rREST provides the optimal interpolating function that can be used when the reference channel is missing or the 'bad' channels are rejected. The derivations and properties imply that: (a) any two URs can transform to each other and referencing with URs multiple times will not accumulate artifacts; (b) whatever URs the EEG data was previously transformed with, the minimum norm solution to the reference problem will be REST and AR with and without modeling volume conduction, respectively; (c) the MLE and the Bayesian theory show the theoretical optimality of REST. The advantages and limitations of AR and REST are discussed to guide readers for their proper use.

Which Reference Should We Use for EEG and ERP practice?

Which reference is appropriate for the scalp ERP and EEG studies? This unsettled problem still inspires unceasing debate. The ideal reference should be the one with zero or constant potential but unfortunately it is well known that no point on the body fulfills this condition. Consequently, more than ten references are used in the present EEG-ERP studies. This diversity seriously undermines the reproducibility and comparability of results across laboratories. A comprehensive review accompanied by a brief communication with rigorous derivations and notable properties (Hu et al. Brain Topogr, 2019. https://doi.org/10.1007/s10548-019-00706-y ) is thus necessary to provide application-oriented principled recommendations. In this paper current popular references are classified into two categories: (1) unipolar references that construct a neutral reference, including both online unipolar references and offline re-references. Examples of unipolar references are the reference electrode standardization technique (REST), average reference (AR), and linked-mastoids/ears reference (LM); (2) non-unipolar references that include the bipolar reference and the Laplacian reference. We show that each reference is derived with a different assumption and serves different aims. We also note from (Hu et al. 2019) that there is a general form for the reference problem, the 'no memory' property of the unipolar references, and a unified estimator for the potentials at infinity termed as the regularized REST (rREST) which has more advantageous statistical evidence than AR. A thorough discussion of the advantages and limitations of references is provided with recommendations in the hope to clarify the role of each reference in the ERP and EEG practice.

[Formula: see text]-[Formula: see text]: A Nonparametric Model for Neural Power Spectra Decomposition.

The power spectra estimated from the brain recordings are the mixed representation of aperiodic transient activity and periodic oscillations, i.e., aperiodic component (AC) and periodic component (PC). Quantitative neurophysiology requires precise decomposition preceding parameterizing each component. However, the shape, statistical distribution, scale, and mixing mechanism of AC and PCs are unclear, challenging the effectiveness of current popular parametric models such as FOOOF, IRASA, BOSC, etc. Here, ξ- π was proposed to decompose the neural spectra by embedding the nonparametric spectra estimation with penalized Whittle likelihood and the shape language modeling into the expectation maximization framework. ξ- π was validated on the synthesized spectra with loss statistics and on the sleep EEG and the large sample iEEG with evaluation metrics and neurophysiological evidence. Compared to FOOOF, both the simulation presenting shape irregularities and the batch simulation with multiple isolated peaks indicated that ξ- π improved the fit of AC and PCs with less loss and higher F1-score in recognizing the centering frequencies and the number of peaks; the sleep EEG revealed that ξ- π produced more distinguishable AC exponents and improved the sleep state classification accuracy; the iEEG showed that ξ- π approached the clinical findings in peak discovery. Overall, ξ- π offered good performance in the spectra decomposition, which allows flexible parameterization using descriptive statistics or kernel functions. ξ- π is a seminal tool for brain signal decoding in fields such as cognitive neuroscience, brain-computer interface, neurofeedback, and brain diseases.

Spatial-frequency-temporal convolutional recurrent network for olfactory-enhanced EEG emotion recognition.

BACKGROUND: Multimedia stimulation of brain activity is important for emotion induction. Based on brain activity, emotion recognition using EEG signals has become a hot issue in the field of affective computing. NEW METHOD: In this paper, we develop a noval odor-video elicited physiological signal database (OVPD), in which we collect the EEG signals from eight participants in positive, neutral and negative emotional states when they are stimulated by synchronizing traditional video content with the odors. To make full use of the EEG features from different domains, we design a 3DCNN-BiLSTM model combining convolutional neural network (CNN) and bidirectional long short term memory (BiLSTM) for EEG emotion recognition. First, we transform EEG signals into 4D representations that retain spatial, frequency and temporal information. Then, the representations are fed into the 3DCNN-BiLSTM model to recognize emotions. CNN is applied to learn spatial and frequency information from the 4D representations. BiLSTM is designed to extract forward and backward temporal dependences. RESULTS: We conduct 5-fold cross validation experiments five times on the OVPD dataset to evaluate the performance of the model. The experimental results show that our presented model achieves an average accuracy of 98.29% with the standard deviation of 0.72% under the olfactory-enhanced video stimuli, and an average accuracy of 98.03% with the standard deviation of 0.73% under the traditional video stimuli on the OVPD dataset in the three-class classification of positive, neutral and negative emotions. To verify the generalisability of our proposed model, we also evaluate this approach on the public EEG emotion dataset (SEED). COMPARISON WITH EXISTING METHOD: Compared with other baseline methods, our designed model achieves better recognition performance on the OVPD dataset. The average accuracy of positive, neutral and negative emotions is better in response to the olfactory-enhanced videos than the pure videos for the 3DCNN-BiLSTM model and other baseline methods. CONCLUSION: The proposed 3DCNN-BiLSTM model is effective by fusing the spatial-frequency-temporal features of EEG signals for emotion recognition. The provided olfactory stimuli can induce stronger emotions than traditional video stimuli and improve the accuracy of emotion recognition to a certain extent. However, superimposing odors unrelated to the video scenes may distract participants' attention, and thus reduce the final accuracy of EEG emotion recognition.

A novel deep learning model based on the ICA and Riemannian manifold for EEG-based emotion recognition.

BACKGROUND: The EEG-based emotion recognition is one of the primary research orientations in the field of emotional intelligence and human-computer interaction (HCI). NEW METHOD: We proposed a novel model, denoted as ICRM-LSTM, for EEG-based emotion recognition by combining the independent component analysis (ICA), the Riemannian manifold (RM), and the long short-term memory network (LSTM). The SEED and MAHNOB-HCI dataset were employed to verify the performance of the proposed model. Firstly, ICA was used to perform blind source separation (BSS) for the preprocessed EEG signals provided by the two datasets. Then, a series of the covariance matrices according to time order were extracted from the blind source signals, and the symmetric positive definiteness of covariance matrix allowed us to project them from RM space to Euclid space by logarithmic mapping. Finally, the transformed covariance matrices were taken as inputs of the LSTM network to perform the emotion recognition. RESULTS: To validate the effect of the ICRM method on the performance of the proposed model, we designed three groups of comparative experiments. The average accuracy of the ICRM-LSTM model were 96.75 % and 98.09 % in SEED and MAHNOB-HCI, respectively. Then we compared our model with the models didn't perform the ICA or RM method, where we found that the proposed model had better performance. Furthermore, to verify the robustness, we added three groups of Gaussian noise with different signal-to-noise ratios (SNR) to the preprocessed EEG signals, and the proposed model achieved a good performance in both the two datasets. COMPARISON WITH EXISTING METHOD: The performance of our model was superior to most of existing methods which also employed the SEED or MAHNOB-HCI dataset. CONCLUSION: This paper demonstrated that the ICA and RM were helpful for EEG-based emotion recognition, and provided the evidence that the RM method could effectively alleviate the problem of the uncertain ordering of ICA.

Stepwise Covariance-Free Common Principal Components (CF-CPC) With an Application to Neuroscience.

Finding the common principal component (CPC) for ultra-high dimensional data is a multivariate technique used to discover the latent structure of covariance matrices of shared variables measured in two or more k conditions. Common eigenvectors are assumed for the covariance matrix of all conditions, only the eigenvalues being specific to each condition. Stepwise CPC computes a limited number of these CPCs, as the name indicates, sequentially and is, therefore, less time-consuming. This method becomes unfeasible when the number of variables p is ultra-high since storing k covariance matrices requires O(k p 2) memory. Many dimensionality reduction algorithms have been improved to avoid explicit covariance calculation and storage (covariance-free). Here we propose a covariance-free stepwise CPC, which only requires O(k n) memory, where n is the total number of examples. Thus for n < < p, the new algorithm shows apparent advantages. It computes components quickly, with low consumption of machine resources. We validate our method CFCPC with the classical Iris data. We then show that CFCPC allows extracting the shared anatomical structure of EEG and MEG source spectra across a frequency range of 0.01-40 Hz.

Movement Symmetry Assessment by Bilateral Motion Data Fusion.

OBJECTIVE: A new approach, named bilateral motion data fusion, was proposed for the analysis of movement symmetry, which takes advantage of cross-information between both sides of the body and processes the unilateral motion data at the same time. METHODS: This was accomplished using canonical correlation analysis and joint independent component analysis. It should be noted that human movements include many categories, which cannot be enumerated one by one. Therefore, the gait rhythm fluctuations of the healthy subjects and patients with neurodegenerative diseases were employed as an example for method illustration. In addition, our model explains the movement data by latent parameters in the time and frequency domains, respectively, which were both based on bilateral motion data fusion. RESULTS: They show that our method not only reflects the physiological correlates of movement but also obtains the differential signatures of movement asymmetry in diverse neurodegenerative diseases. Furthermore, the latent variables also exhibit the potentials for sharper disease distinctions. CONCLUSION: We have provided a new perspective on movement analysis, which may prove to be a promising approach. SIGNIFICANCE: This method exhibits the potentials for effective movement feature extractions, which might contribute to many research fields such as rehabilitation, neuroscience, biomechanics, and kinesiology.

Unified Bayesian Estimator of EEG Reference at Infinity: rREST (Regularized Reference Electrode Standardization Technique).

The choice of reference for the electroencephalogram (EEG) is a long-lasting unsolved issue resulting in inconsistent usages and endless debates. Currently, both the average reference (AR) and the reference electrode standardization technique (REST) are two primary, apparently irreconcilable contenders. We propose a theoretical framework to resolve this reference issue by formulating both (a) estimation of potentials at infinity, and (b) determination of the reference, as a unified Bayesian linear inverse problem, which can be solved by maximum a posterior estimation. We find that AR and REST are very particular cases of this unified framework: AR results from biophysically non-informative prior; while REST utilizes the prior based on the EEG generative model. To allow for simultaneous denoising and reference estimation, we develop the regularized versions of AR and REST, named rAR and rREST, respectively. Both depend on a regularization parameter that is the noise to signal variance ratio. Traditional and new estimators are evaluated with this framework, by both simulations and analysis of real resting EEGs. Toward this end, we leverage the MRI and EEG data from 89 subjects which participated in the Cuban Human Brain Mapping Project. Generated artificial EEGs-with a known ground truth, show that relative error in estimating the EEG potentials at infinity is lowest for rREST. It also reveals that realistic volume conductor models improve the performances of REST and rREST. Importantly, for practical applications, it is shown that an average lead field gives the results comparable to the individual lead field. Finally, it is shown that the selection of the regularization parameter with Generalized Cross-Validation (GCV) is close to the "oracle" choice based on the ground truth. When evaluated with the real 89 resting state EEGs, rREST consistently yields the lowest GCV. This study provides a novel perspective to the EEG reference problem by means of a unified inverse solution framework. It may allow additional principled theoretical formulations and numerical evaluation of performance.

How do reference montage and electrodes setup affect the measured scalp EEG potentials?

OBJECTIVE: Human scalp electroencephalogram (EEG) is widely applied in cognitive neuroscience and clinical studies due to its non-invasiveness and ultra-high time resolution. However, the representativeness of the measured EEG potentials for the underneath neural activities is still a problem under debate. This study aims to investigate systematically how both reference montage and electrodes setup affect the accuracy of EEG potentials. APPROACH: First, the standard EEG potentials are generated by the forward calculation with a single dipole in the neural source space, for eleven channel numbers (10, 16, 21, 32, 64, 85, 96, 128, 129, 257, 335). Here, the reference is the ideal infinity implicitly determined by forward theory. Then, the standard EEG potentials are transformed to recordings with different references including five mono-polar references (Left earlobe, Fz, Pz, Oz, Cz), and three re-references (linked mastoids (LM), average reference (AR) and reference electrode standardization technique (REST)). Finally, the relative errors between the standard EEG potentials and the transformed ones are evaluated in terms of channel number, scalp regions, electrodes layout, dipole source position and orientation, as well as sensor noise and head model. MAIN RESULTS: Mono-polar reference recordings are usually of large distortions; thus, a re-reference after online mono-polar recording should be adopted in general to mitigate this effect. Among the three re-references, REST is generally superior to AR for all factors compared, and LM performs worst. REST is insensitive to head model perturbation. AR is subject to electrodes coverage and dipole orientation but no close relation with channel number. SIGNIFICANCE: These results indicate that REST would be the first choice of re-reference and AR may be an alternative option for high level sensor noise case. Our findings may provide the helpful suggestions on how to obtain the EEG potentials as accurately as possible for cognitive neuroscientists and clinicians.

How do the reference montage and electrodes setup affect the measured scalp EEG potentials?

OBJECTIVE: Human scalp electroencephalogram (EEG) is widely applied in cognitive neuroscience and clinical studies due to its non-invasiveness and ultra-high time resolution. However, the representativeness of the measured EEG potentials for the underneath neural activities is still a problem under debate. This study aims to investigate systematically how both reference montage and electrodes setup affect the accuracy of EEG potentials. APPROACH: First, the standard EEG potentials are generated by the forward calculation with a single dipole in the neural source space, for eleven channel numbers (10, 16, 21, 32, 64, 85, 96, 128, 129, 257, 335). Here, the reference is the ideal infinity implicitly determined by forward theory. Then, the standard EEG potentials are transformed to recordings with different references including five monopolar references (Left earlobe, Fz, Pz, Oz, Cz), and three re-references (Linked Mastoids (LM), Average Reference (AR) and Reference Electrode Standardization Technique (REST)). Finally, the relative errors between the standard EEG potentials and the transformed ones are evaluated in terms of channel number, scalp regions, electrodes layout, dipole source position and orientation, as well as sensor noise and head model. MAIN RESULTS: Mono-polar reference recordings are usually of large distortions; thus, a re-reference after online mono-polar recording should be adopted in general to mitigate this effect. Among the three re-references, REST is generally superior to AR for all factors compared, and LM performs worst. REST is insensitive to head model perturbation. AR is subject to electrodes coverage and dipole orientation but no close relation with channel number. SIGNIFICANCE: These results indicate that REST would be the first choice of re-reference and AR may be an alternative option for high level sensor noise case. Our findings may provide the helpful suggestions on how to obtain the EEG potentials as accurately as possible for cognitive neuroscientists and clinicians.