Hybrid ICA-Regression: Automatic Identification and Removal of Ocular Artifacts from Electroencephalographic Signals

A Learnable and Explainable Wavelet Neural Network for EEG Artifacts Detection and Classification

Electroencephalography (EEG) artifacts are very common in clinical diagnosis and can heavily impact diagnosis. Manual screening of artifact events is labor-intensive with little benefit. Therefore, exploring algorithms for automatic detection and classification of EEG artifacts can significantly assist clinical diagnosis. In this paper, we propose a learnable and explainable wavelet neural network (WaveNet) for EEG artifact detection and classification. The model is powered by the wavelet decomposition block based on invertible neural network, which can extract signal features without information loss, and a tree generator for building wavelet tree structure automatically. They provide the model with good feature extraction capabilities and explainability. To evaluate the model's performance more fairly, we introduce the base point level matching score (BASE) and the Event-Aligned Compensation Scoring (EACS) at the event level as two metrics for model performance evaluation. On the challenging Temple University EEG Artifact (TUAR) dataset, our model outperforms other baselines in terms of F1-score for both artifact detection and classification tasks. The case study also validates the model's ability to offer explainability for predictions based on frequency band energy, suggesting potential applications in clinical diagnosis.

Correlation-Filter-Based Channel and Feature Selection Framework for Hybrid EEG-fNIRS BCI Applications

The proposed study is based on a feature and channel selection strategy that uses correlation filters for brain-computer interface (BCI) applications using electroencephalography (EEG)-functional near-infrared spectroscopy (fNIRS) brain imaging modalities. The proposed approach fuses the complementary information of the two modalities to train the classifier. The channels most closely correlated with brain activity are extracted using a correlation-based connectivity matrix for fNIRS and EEG separately. Furthermore, the training vector is formed through the identification and fusion of the statistical features of both modalities (i.e., slope, skewness, maximum, skewness, mean, and kurtosis). The constructed fused feature vector is passed through various filters (including ReliefF, minimum redundancy maximum relevance, chi-square test, analysis of variance, and Kruskal-Wallis filters) to remove redundant information before training. Traditional classifiers such as neural networks, support-vector machines, linear discriminant analysis, and ensembles were used for the purpose of training and testing. A publicly available dataset with motor imagery information was used for validation of the proposed approach. Our findings indicate that the proposed correlation-filter-based channel and feature selection framework significantly enhances the classification accuracy of hybrid EEG-fNIRS. The ReliefF-based filter outperformed other filters with the ensemble classifier with a high accuracy of 94.77 ± 4.26%. The statistical analysis also validated the significance (p < 0.01) of the results. A comparison of the proposed framework with the prior findings was also presented. Our results show that the proposed approach can be used in future EEG-fNIRS-based hybrid BCI applications.

Multiresolution directed transfer function approach for segment-wise seizure classification of epileptic EEG signal

Currently, with the bloom in artificial intelligence (AI) algorithms, various human-centered smart systems can be utilized, especially in cognitive computing, for the detection of various chronic brain diseases such as epileptic seizure. The primary goal of this research article is to propose a novel human-centered cognitive computing (HCCC) method for segment-wise seizure classification by employing multiresolution extracted data with directed transfer function (DTF) features, termed as the multiresolution directed transfer function (MDTF) approach. Initially, the multiresolution information of the epileptic seizure signal is extracted using a multiresolution adaptive filtering (MRAF) method. These seizure details are passed to the DTF where the information flow of high frequency bands is computed. Thereafter, different measures of complexity such as approximate entropy (AEN) and sample entropy (SAEN) are computed from the extracted high frequency bands. Lastly, a k-nearest neighbor (k-NN) and support vector machine (SVM) are used for classifying the EEG signal into non-seizure and seizure data depending on the multiresolution based information flow characteristics. The MDTF approach is tested on a standard dataset and validated using a dataset from a local hospital. The proposed technique has obtained an average sensitivity of 98.31%, specificity of 96.13% and accuracy of 98.89% using SVM classifier. The average detection rate of the MDTF approach is 97.72% which is greater than the existing approaches. The proposed MDTF method will help neuro-specialists to locate seizure information drift which occurs within the consecutive segments and between two channels. The main advantage of the MDTF approach is its capability to locate the seizure activity contained by the EEG signal with accuracy. This will assist the neurologists with the precise localization of the epileptic seizure automatically and hence will reduce the burden of time-consuming epileptic seizure analysis.

Multi-Algorithm Artifact Correction (MAAC) procedure part one: Algorithm and example

The Multi-Algorithm Artifact Correction (MAAC) procedure is presented for electroencephalographic (EEG) data, as made freely available in the open-source EP Toolkit (Dien, 2010). First the major EEG artifact correction methods (regression, spatial filters, principal components analysis, and independent components analysis) are reviewed. Contrary to the dominant approach of picking one method that is thought to be most effective, this review concludes that none are globally superior, but rather each has strengths and weaknesses. Then each of the major artifact types are reviewed (Blink, Corneo-Retinal Dipole, Saccadic Spike Potential, and Movement). For each one, it is proposed that one of the major correction methods is best matched to address it, resulting in the MAAC procedure. The MAAC itself is then presented, as implemented in the EP Toolkit, in order to provide a sense of the user experience. The primary goal of this present paper is to make the conceptual argument for the MAAC approach.

Deep insights into MCI diagnosis: A comparative deep learning analysis of EEG time series

BACKGROUND

Individuals in the early stages of Alzheimer's Disease (AD) are typically diagnosed with Mild Cognitive Impairment (MCI). MCI represents a transitional phase between normal cognitive function and AD. Electroencephalography (EEG) records carry valuable insights into cerebral cortex brain activities to analyze neuronal degeneration. To enhance the precision of dementia diagnosis, automatic and intelligent methods are required for the analysis and processing of EEG signals.

NEW METHODS

This paper aims to address the challenges associated with MCI diagnosis by leveraging EEG signals and deep learning techniques. The analysis in this study focuses on processing the information embedded within the sequence of raw EEG time series data. EEG recordings are collected from 10 Healthy Controls (HC) and 10 MCI participants using 19 electrodes during a 30 min eyes-closed session. EEG time series are transformed into 2 separate formats of input tensors and applied to deep neural network architectures. Convolutional Neural Network (CNN) and ResNet from scratch are performed with 2D time series with different segment lengths. Furthermore, EEGNet and DeepConvNet architectures are utilized for 1D time series.

RESULTS

ResNet demonstrates superior effectiveness in detecting MCI when compared to CNN architecture. Complete discrimination is achieved using EEGNet and DeepConvNet for noisy segments.

COMPARISON WITH EXISTING METHODS

ResNet has yielded a 3 % higher accuracy rate compared to CNN. None of the architectures in the literature have achieved 100 % accuracy except proposed EEGNet and DeepConvnet.

CONCLUSION

Deep learning architectures hold great promise in enhancing the accuracy of early MCI detection.

State-of-the-Art on Brain-Computer Interface Technology

This paper provides a comprehensive overview of the state-of-the-art in brain-computer interfaces (BCI). It begins by providing an introduction to BCIs, describing their main operation principles and most widely used platforms. The paper then examines the various components of a BCI system, such as hardware, software, and signal processing algorithms. Finally, it looks at current trends in research related to BCI use for medical, educational, and other purposes, as well as potential future applications of this technology. The paper concludes by highlighting some key challenges that still need to be addressed before widespread adoption can occur. By presenting an up-to-date assessment of the state-of-the-art in BCI technology, this paper will provide valuable insight into where this field is heading in terms of progress and innovation.

OptEF-BCI: An Optimization-Based Hybrid EEG and fNIRS-Brain Computer Interface

Multimodal data fusion (electroencephalography (EEG) and functional near-infrared spectroscopy (fNIRS)) has been developed as an important neuroimaging research field in order to circumvent the inherent limitations of individual modalities by combining complementary information from other modalities. This study employed an optimization-based feature selection algorithm to systematically investigate the complementary nature of multimodal fused features. After preprocessing the acquired data of both modalities (i.e., EEG and fNIRS), the temporal statistical features were computed separately with a 10 s interval for each modality. The computed features were fused to create a training vector. A wrapper-based binary enhanced whale optimization algorithm (E-WOA) was used to select the optimal/efficient fused feature subset using the support-vector-machine-based cost function. An online dataset of 29 healthy individuals was used to evaluate the performance of the proposed methodology. The findings suggest that the proposed approach enhances the classification performance by evaluating the degree of complementarity between characteristics and selecting the most efficient fused subset. The binary E-WOA feature selection approach showed a high classification rate (94.22 ± 5.39%). The classification performance exhibited a 3.85% increase compared with the conventional whale optimization algorithm. The proposed hybrid classification framework outperformed both the individual modalities and traditional feature selection classification (p < 0.01). These findings indicate the potential efficacy of the proposed framework for several neuroclinical applications.

Traumatic Brain Injury (TBI) Detection: Past, Present, and Future

Traumatic brain injury (TBI) can produce temporary biochemical imbalance due to leaks through cell membranes or disruption of the axoplasmic flow due to the misalignment of intracellular neurofilaments. If untreated, TBI can lead to Alzheimer's, Parkinson's, or total disability. Mild TBI (mTBI) accounts for about about 90 percent of all TBI cases. The detection of TBI as soon as it happens is crucial for successful treatment management. Neuroimaging-based tests provide only a structural and functional mapping of the brain with poor temporal resolution. Such tests may not detect mTBI. On the other hand, the electroencephalogram (EEG) provides good spatial resolution and excellent temporal resolution of the brain activities beside its portability and low cost. The objective of this paper is to provide clinicians and scientists with a one-stop source of information to quickly learn about the different technologies used for TBI detection, their advantages and limitations. Our research led us to conclude that even though EEG-based TBI detection is potentially a powerful technology, it is currently not able to detect the presence of a mTBI with high confidence. The focus of the paper is to review existing approaches and provide the reason for the unsuccessful state of EEG-based detection of mTBI.

Single channel approach for filtering electroencephalographic signals strongly contaminated with facial electromyography

Eliminating facial electromyographic (EMG) signal from the electroencephalogram (EEG) is crucial for the accuracy of applications such as brain computer interfaces (BCIs) and brain functionality measurement. Facial electromyography typically corrupts the electroencephalogram. Although it is possible to find in the literature a number of multi-channel approaches for filtering corrupted EEG, studies employing single-channel approaches are scarce. In this context, this study proposed a single-channel method for attenuating facial EMG noise from contaminated EEG. The architecture of the method allows for the evaluation and incorporation of multiple decomposition and adaptive filtering techniques. The decomposition method was responsible for generating EEG or EMG reference signals for the adaptive filtering stage. In this study, the decomposition techniques CiSSA, EMD, EEMD, EMD-PCA, SSA, and Wavelet were evaluated. The adaptive filtering methods RLS, Wiener, LMS, and NLMS were investigated. A time and frequency domain set of features were estimated from experimental signals to evaluate the performance of the single channel method. This set of characteristics permitted the characterization of the contamination of distinct facial muscles, namely Masseter, Frontalis, Zygomatic, Orbicularis Oris, and Orbicularis Oculi. Data were collected from ten healthy subjects executing an experimental protocol that introduced the necessary variability to evaluate the filtering performance. The largest level of contamination was produced by the Masseter muscle, as determined by statistical analysis of the set of features and visualization of topological maps. Regarding the decomposition method, the SSA method allowed for the generation of more suitable reference signals, whereas the RLS and NLMS methods were more suitable when the reference signal was derived from the EEG. In addition, the LMS and RLS methods were more appropriate when the reference signal was the EMG. This study has a number of practical implications, including the use of filtering techniques to reduce EEG contamination caused by the activation of facial muscles required by distinct types of studies. All the developed code, including examples, is available to facilitate a more accurate reproduction and improvement of the results of this study.

Intelligent Method for Real-Time Portable EEG Artifact Annotation in Semiconstrained Environment Based on Computer Vision

As a convenient device for observing neural activity in the natural environment, portable EEG technology (PEEGT) has an extensive prospect in expanding neuroscience research into natural applications. However, unlike in the laboratory environment, PEEGT is usually applied in a semiconstrained environment, including management and engineering, generating much more artifacts caused by the subjects' activities. Due to the limitations of existing artifacts annotation, the problem limits PEEGT to take advantage of portability and low-test cost, which is a crucial obstacle for the potential application of PEEGT in the natural environment. This paper proposes an intelligent method to identify two leading antecedent causes of EEG artifacts, participant's blinks and head movements, and annotate the time segments of artifacts in real time based on computer vision (CV). Furthermore, it changes the original postprocessing mode based on artifact signal recognition to the preprocessing mode based on artifact behavior recognition by the CV method. Through a comparative experiment with three artifacts mark operators and the CV method, we verify the effectiveness of the method, which lays a foundation for accurate artifact removal in real time in the next step. It enlightens us on how to adopt computer technology to conduct large-scale neurotesting in a natural semiconstrained environment outside the laboratory without expensive laboratory equipment or high manual costs.

Complexity of EEG Dynamics for Early Diagnosis of Alzheimer's Disease Using Permutation Entropy Neuromarker

BACKGROUND AND OBJECTIVE

Electroencephalogram (EEG) is one of the most demanded screening tools that investigates the effects of Alzheimer's Disease (AD) on human brain. Identification of AD in early stage gives rise to efficient treatment in dementia. Mild Cognitive Impairment (MCI) is considered as a conversion stage. Reducing EEG complexity can be used as a marker to detect AD. The aim of this study is to develop a 3-way diagnostic classification using EEG complexity in the detection of MCI/AD in clinical practice. This study also investigates the effects of different eyes states, i.e. eyes-open, eyes-closed on classification performance.

METHODS

EEG recordings from 85 AD, 85 MCI subjects, and 85 Healthy Controls with eyes-open and eyes- closed are analyzed. Permutation Entropy (PE) values are computed from frontal, central, parietal, temporal, and occipital regions for each EEG epoch. Distribution of PE values are visualized to observe discrimination of MCI/AD with HC. Visual investigations are combined with statistical analysis using ANOVA to determine whether groups are significant or not. Multinomial Logistic Regression model is applied to feature sets in order to classify participants individually.

RESULTS

Distribution of measured PE shows that EEG complexity is lower in AD and higher in HC group. MCI group is observed as an intermediate form due to heterogeneous values. Results from 3-way classification indicate that F1-scores and rates of sensitivity and specificity achieve the highest overall discrimination rates reaching up to 100% for at TP8 for eyes-closed condition; and C3, C4, T8, O2 electrodes for eyes-open condition. Classification of HC from both patient groups is achieved best. Eyes-open state increases discrimination of MCI and AD.

CONCLUSIONS

This nonlinear EEG methodology study contributes to literature with high discrimination rates for identification of AD. PE is recommended as a practical diagnostic neuro-marker for AD studies. Resting state EEG at eyes-open condition can be more advantageous over eyes-closed EEG recordings for diagnosis of AD.

Lag synchronization of coupled time-delayed FitzHugh-Nagumo neural networks via feedback control

Synchronization plays a significant role in information transfer and decision-making by neurons and brain neural networks. The development of control strategies for synchronizing a network of chaotic neurons with time delays, different direction-dependent coupling (unidirectional and bidirectional), and noise, particularly under external disturbances, is an essential and very challenging task. Researchers have extensively studied the synchronization mechanism of two coupled time-delayed neurons with bidirectional coupling and without incorporating the effect of noise, but not for time-delayed neural networks. To overcome these limitations, this study investigates the synchronization problem in a network of coupled FitzHugh-Nagumo (FHN) neurons by incorporating time delays, different direction-dependent coupling (unidirectional and bidirectional), noise, and ionic and external disturbances in the mathematical models. More specifically, this study investigates the synchronization of time-delayed unidirectional and bidirectional ring-structured FHN neuronal systems with and without external noise. Different gap junctions and delay parameters are used to incorporate time-delay dynamics in both neuronal networks. We also investigate the influence of the time delays between connected neurons on synchronization conditions. Further, to ensure the synchronization of the time-delayed FHN neuronal networks, different adaptive control laws are proposed for both unidirectional and bidirectional neuronal networks. In addition, necessary and sufficient conditions to achieve synchronization are provided by employing the Lyapunov stability theory. The results of numerical simulations conducted for different-sized multiple networks of time-delayed FHN neurons verify the effectiveness of the proposed adaptive control schemes.

Validation of SOBI-DANS method for automatic identification of horizontal and vertical eye movement components from EEG

Neurophysiological investigation of neural processes are hindered by the presence of large artifacts associated with eye movement. Although blind source separation (BSS)-based hybrid algorithms are useful for separating, identifying, and removing these artifacts from EEG, it remains unexamined to what extent neural signals can remain mixed with these artifact components, potentially resulting in unintended removal of critical neural signals. Here, we present a novel validation approach to quantitatively evaluate to what extent horizontal and vertical saccadic eye movement-related artifact components (H and V Comps) are indeed ocular in origin. To automate the identification of the H and V Comps recovered by the second-order blind identification (SOBI), we introduced a novel Discriminant ANd Similarity (DANS)-based method. Through source localization, we showed that over 95% of variance in the SOBI-DANS identified H and V Comps' scalp projections were ocular in origin. Through the analysis of saccade-related potentials (SRPs), we found that the H and V Comps' SRP amplitudes were finely modulated by eye movement direction and distance jointly. SOBI-DANS' component selection was in 100% agreement with human experts' selection and was 100% successful in component identification across all participants indicating a high cross-individual consistency or robustness. These results set the stage for future work to transform the to-be-thrown-away artifacts into signals indicative of gaze position, thereby providing readily co-registered eye movement and neural signal without using a separate eye tracker.

Improving EEG Muscle Artifact Removal With an EMG Array

Removal of artifacts induced by muscle activity is crucial for analysis of the electroencephalogram (EEG), and continues to be a challenge in experiments where the subject may speak, change facial expressions, or move. Ensemble empirical mode decomposition with canonical correlation analysis (EEMD-CCA) has been proven to be an efficient method for denoising of EEG contaminated with muscle artifacts. EEMD-CCA, likewise the majority of algorithms, does not incorporate any statistical information of the artifact, namely, electromyogram (EMG) recorded over the muscles actively contaminating the EEG. In this paper, we propose to extend EEMD-CCA in order to include an EMG array as information to aid the removal of artifacts, assessing the performance gain achieved when the number of EMG channels grow. By filtering adaptively (recursive least squares, EMG array as reference) each component resulting from CCA, we aim to ameliorate the distortion of brain signals induced by artifacts and denoising methods. We simulated several noise scenarios based on a linear contamination model, between real and synthetic EEG and EMG signals, and varied the number of EMG channels available to the filter. Our results exhibit a substantial improvement in the performance as the number of EMG electrodes increase from 2 to 16. Further increasing the number of EMG channels up to 128 did not have a significant impact on the performance. We conclude by recommending the use of EMG electrodes to filter components, as it is a computationally inexpensive enhancement that impacts significantly on performance using only a few electrodes.

A Hybrid Speller Design Using Eye Tracking and SSVEP Brain-Computer Interface

Steady-state visual evoked potentials (SSVEPs) have been extensively utilized to develop brain-computer interfaces (BCIs) due to the advantages of robustness, large number of commands, high classification accuracies, and information transfer rates (ITRs). However, the use of several simultaneous flickering stimuli often causes high levels of user discomfort, tiredness, annoyingness, and fatigue. Here we propose to design a stimuli-responsive hybrid speller by using electroencephalography (EEG) and video-based eye-tracking to increase user comfortability levels when presented with large numbers of simultaneously flickering stimuli. Interestingly, a canonical correlation analysis (CCA)-based framework was useful to identify target frequency with a 1 s duration of flickering signal. Our proposed BCI-speller uses only six frequencies to classify forty-eight targets, thus achieve greatly increased ITR, whereas basic SSVEP BCI-spellers use an equal number of frequencies to the number of targets. Using this speller, we obtained an average classification accuracy of 90.35 ± 3.597% with an average ITR of 184.06 ± 12.761 bits per minute in a cued-spelling task and an ITR of 190.73 ± 17.849 bits per minute in a free-spelling task. Consequently, our proposed speller is superior to the other spellers in terms of targets classified, classification accuracy, and ITR, while producing less fatigue, annoyingness, tiredness and discomfort. Together, our proposed hybrid eye tracking and SSVEP BCI-based system will ultimately enable a truly high-speed communication channel.

A Method for Tracking the Time Evolution of Steady-State Evoked Potentials

Neural entrainment refers to the synchronization of neural activity to the periodicity of sensory stimuli. This synchronization defines the generation of steady-state evoked responses (i.e., oscillations in the electroencephalogram phase-locked to the driving stimuli). The classic interpretation of the amplitude of the steady-state evoked responses assumes a stereotypical time-invariant neural response plus random background fluctuations, such that averaging over repeated presentations of the stimulus recovers the stereotypical response. This approach ignores the dynamics of the steady-state, as in the case of the adaptation elicited by prolonged exposures to the stimulus. To analyze the dynamics of steady-state responses, it can be assumed that the time evolution of the response amplitude is the same in different stimulation runs separated by sufficiently long breaks. Based on this assumption, a method to characterize the time evolution of steady-state responses is presented. A sufficiently large number of recordings are acquired in response to the same experimental condition. Experimental runs (recordings) are column-wise averaged (i.e., runs are averaged but epoch within recordings are not averaged with the preceding segments). The column-wise averaging allows analysis of steady-state responses in recordings with remarkably high signal-to-noise ratios. Therefore, the averaged signal provides an accurate representation of the time evolution of the steady-state response, which can be analyzed in both the time and frequency domains. In this study, a detailed description of the method is provided, using steady-state visually evoked potentials as an example of a response. Advantages and caveats are evaluated based on a comparison with single-trial methods designed to analyze neural entrainment.

A Novel Artifact Reconstruction Method Applied to Blood Pressure Signals

Physiological records are one of the most relevant elements to obtain objective information from the patients. The presence of artifacts in biomedical signals can give misleading in the analysis of information that these signals give. The blood pressure signal is one of the records clearly affected by different artifacts, especially the ones due from the calibration episodes. We propose a method to reconstruct different episodes of artifacts in these signals. This method is sustained on the detection of the events of the signal, differentiating between to the physiological cycles and the artifacts. The performance of the method is based on the detection of the cycles and artifact's position, the identification of the number of cycles to reconstruct, and the prediction of the cycle model used to generate the missing cycles. The parameter θ_E represents the difference between the area under the curve when two events are compared. The value of this parameter is low when two similar events are compared like the physiological cycles, whereas it is high comparing a cycle with an artifact. An adaptive threshold is defined to identify the artifact episodes. The number of cycles to reconstruct is generated considering the same number of their neighbours physiological cycles, to left and right, of the original signal. Finally, the performance of the method has been analyzed comparing the number of events and artifacts detected and their correct reconstruction. According to the results, the reconstruction error was less than 1% in all cases.

Differential Path-Length Factor's Effect on the Characterization of Brain's Hemodynamic Response Function: A Functional Near-Infrared Study

Functional near-infrared spectroscopy (fNIRS) has evolved as a neuro-imaging modality over the course of the past two decades. The removal of superfluous information accompanying the optical signal, however, remains a challenge. A comprehensive analysis of each step is necessary to ensure the extraction of actual information from measured fNIRS waveforms. A slight change in shape could alter the features required for fNIRS-BCI applications. In the present study, the effect of the differential path-length factor (DPF) values on the characteristics of the hemodynamic response function (HRF) was investigated. Results were compiled for both simulated data sets and healthy human subjects over a range of DPF values from three to eight. Different sets of activation durations and stimuli were used to generate the simulated signals for further analysis. These signals were split into optical densities under a constrained environment utilizing known values of DPF. Later, different values of DPF were used to analyze the variations of actual HRF. The results, as summarized into four categories, suggest that the DPF can change the main and post-stimuli responses in addition to other interferences. Six healthy subjects participated in this study. Their observed optical brain time-series were fed into an iterative optimization problem in order to estimate the best possible fit of HRF and physiological noises present in the measured signals with free parameters. A series of solutions was derived for different values of DPF in order to analyze the variations of HRF. It was observed that DPF change is responsible for HRF creep from actual values as well as changes in HRF characteristics.

Method for Improving EEG Based Emotion Recognition by Combining It with Synchronized Biometric and Eye Tracking Technologies in a Non-invasive and Low Cost Way

Technical advances, particularly the integration of wearable and embedded sensors, facilitate tracking of physiological responses in a less intrusive way. Currently, there are many devices that allow gathering biometric measurements from human beings, such as EEG Headsets or Health Bracelets. The massive data sets generated by tracking of EEG and physiology may be used, among other things, to infer knowledge about human moods and emotions. Apart from direct biometric signal measurement, eye tracking systems are nowadays capable of determining the point of gaze of the users when interacting in ICT environments, which provides an added value research on many different areas, such as psychology or marketing. We present a process in which devices for eye tracking, biometric, and EEG signal measurements are synchronously used for studying both basic and complex emotions. We selected the least intrusive devices for different signal data collection given the study requirements and cost constraints, so users would behave in the most natural way possible. On the one hand, we have been able to determine basic emotions participants were experiencing by means of valence and arousal. On the other hand, a complex emotion such as empathy has also been detected. To validate the usefulness of this approach, a study involving forty-four people has been carried out, where they were exposed to a series of affective stimuli while their EEG activity, biometric signals, and eye position were synchronously recorded to detect self-regulation. The hypothesis of the work was that people who self-regulated would show significantly different results when analyzing their EEG data. Participants were divided into two groups depending on whether Electro Dermal Activity (EDA) data indicated they self-regulated or not. The comparison of the results obtained using different machine learning algorithms for emotion recognition shows that using EEG activity alone as a predictor for self-regulation does not allow properly determining whether a person in self-regulation its emotions while watching affective stimuli. However, adequately combining different data sources in a synchronous way to detect emotions makes it possible to overcome the limitations of single detection methods.