Marcin Jurczak

Removing artifacts from electroencephalography (EEG) signals is a common technique. Although numerous algorithms have been proposed, most rely solely on EEG data. In this study, we introduce a novel approach utilizing a hybrid convolutional neural network–long short-term memory (CNN-LSTM) architecture alongside simultaneous recording of facial and neck EMG signals. This setup enables the precise elimination of artifacts from the EEG signal. To validate the method, we collected a dataset from 24 participants who were presented with a light-emitting diode (LED) stimulus that elicited steady-state visual evoked potentials (SSVEPs) while they performed strong jaw clenching, an action known to induce significant artifacts. We then assessed the algorithm’s ability to remove artifacts while preserving SSVEP responses. The results were compared against other commonly used algorithms, such as independent component analysis and linear regression. The findings demonstrate that the proposed method exhibits excellent performance, effectively removing artifacts while retaining the EEG signal’s useful components.

The objective of the article is to recognize users’ emotions by classifying facial electromyographic (EMG) signals. A biomedical signal amplifier, equipped with eight active electrodes positioned in accordance with the Facial Action Coding System, was used to record the EMG signals. These signals were registered during a procedure where users acted out various emotions: joy, sadness, surprise, disgust, anger, fear, and neutral. Recordings were made for 16 users. The mean power of the EMG signals formed the feature set. We utilized these features to train and evaluate various classifiers. In the subject-dependent model, the average classification accuracies were 96.3% for KNN, 94.9% for SVM with a linear kernel, 94.6% for SVM with a cubic kernel, and 93.8% for LDA. In the subject-independent model, the classification results varied depending on the tested user, ranging from 91.4% to 48.6% for the KNN classifier, with an average accuracy of 67.5%. The SVM with a cubic kernel performed slightly worse, achieving an average accuracy of 59.1%, followed by the SVM with a linear kernel at 53.9%, and the LDA classifier at 41.2%. Additionally, the study identified the most effective electrodes for distinguishing between pairs of emotions.

Artifacts pose a significant challenge in the analysis of EEG signals. In this study, the authors investigated the impact of artifacts on the detection of steady-state visually evoked potentials (SSVEPs). The article explored various techniques for physiological artifact elimination, including linear regression, adaptive filters, and independent component analysis (ICA). The effectiveness of the algorithms was evaluated using classification accuracy as a metric. The results indicate that the most promising outcomes were achieved with independent component analysis.

Steady-state visually evoked potentials (SSVEPs) are recorded from the visual cortex and are widely used for brain–computer interfaces. Artifacts caused by muscle activity [electromyography (EMG)] and eye movements [electrooculography (EOG)] can reach several millivolts, while brain activity signals [electroencephalography (EEG)] have only several microvolts. EEG signals can be distorted by various types of artifacts. The authors are not aware of any works that have assessed the impact of such artifacts on the registration of SSVEP. In this letter, we examined the influence of different types of artifacts on the power of the EEG signal recorded from the visual cortex. Artifacts were recorded from 12 participants using eight EMG/EOG and eight EEG electrodes. We also identified the locations where it is worth recording EOG/EMG signals to use them to remove artifacts. Our results show that neck strain, swallowing, jaw clenching, and wincing are the most disruptive artifacts for SSVEP.