eeg data processing and analyses  (MathWorks Inc)

Journal: Device

Article Title: Design of an injectable, self-adhesive, and highly stable hydrogel electrode for sleep recording

doi: 10.1016/j.device.2023.100182

Figure Lengend Snippet: Comparison of Long-term EEG signal quality between AIRTrodes and the commercial EEG gel electrodes (A) The time-series EEG signals were recorded from the volunteer by AIRTrode (top) and commercial EEG gel (bottom) during the 100 s of the eyes-open (EO) and eyes-closed (EC) paradigm before overnight sleep. The [1, 30] Hz band-pass filter was applied to remove any direct current (DC) offset. (B) The spectrograms for EEG signals in (A). (C) The PSD analysis of the grand average of all EC and EO epochs for both types of electrodes. The AIRTrode shows an overall lower background signal (frequencies outside of the alpha band [8, 13] Hz) compared with the commercial EEG gel at the specific channel (PO4) before overnight sleep. (D) The mean power spectral density (PSD) analysis derived from the alpha rhythm for the AIRTrode and commercial EEG gel to detect differences in EEG activity between EO and EC periods (****p < 0.0001, n = 5). (E) Signal-to-noise ratio (SNR) comparison between AIRTrodes and commercial EEG gel electrodes. The signal of interest is defined as alpha rhythm, and the background signal is defined as 5–30 Hz, excluding the alpha rhythm. (F) The time-series EEG signals recorded by the AIRTrode (top) and commercial EEG gel (bottom) during the 100 s of the EO-EC paradigm after overnight sleep. The same setting as in (A) was applied. (G) The spectrograms for the EEG signals in (F). Compared with before sleep, the signal quality improved with both types of electrodes. (H) PSD analysis of the grand average of all EC and EO epochs for both types of electrodes. The PSD variation between epochs is greater than before sleep. (I) PSD from the alpha rhythm to show the capability of both types of electrodes to detect differences in EEG activity between EO and EC periods (***p < 0.001, n = 5). (J) SNR comparison between AIRTrodes and commercial EEG gel electrodes after sleep. (***p < 0.001, n = 5).

Article Snippet: All EEG signal data processing, analyses, and visualizations were performed in MATLAB.

Techniques: Comparison, Derivative Assay, Activity Assay

Journal: Hearing research

Article Title: Auditory attention decoding from EEG-based Mandarin speech envelope reconstruction.

doi: 10.1016/j.heares.2022.108552

Figure Lengend Snippet: Fig. 1. AAD framework and speech envelope in continuous Mandarin speech. (a) Speech envelopes for two example sentences (transcript in Chinese and Pinyin) spoken by a native Mandarin speaker. (b) AAD framework for Mandarin speech. According to the oscillation rhythm, EEG signals were divided into the low-frequency band (0.1–15 Hz) and beta-band (15–30 Hz). We used the low-frequency band and whole rhythm (0.1–30 Hz) for AAD. 60-channel of the whole-brain and optimized 17-channel (mainly including STG and Prefrontal lobe) EEG data were used to reconstruct the speech envelope based on the LSTM model. The correlation analysis was carried out according to the reconstructed envelope and the speech envelope of multiple speech sources. The most relevant speech was determined as the subject of attention.

Article Snippet: Our work chieved an accuracy of 74.29% within a 0.15 s decision window, hich may provide a strong practical application in hearing aids. urthermore, the architecture could help develop neuro-steered earing devices, especially those who speak tonal languages. ata and code availability EEG data analyses were performed in MATLAB, the freely availble toolbox EEGLAB, and Python 3.8.

Techniques: