CLEEGN: A Convolutional Neural Network for Plug-and-Play Automatic EEG Reconstruction

TL;DR

The results of simulated online validation suggest that, even without any calibration, CLEEGN can largely preserve inherent brain activity and outperforms leading online/offline artifact removal methods in the decoding accuracy of reconstructed EEG data.

Abstract

Human electroencephalography (EEG) is a brain monitoring modality that senses cortical neuroelectrophysiological activity in high-temporal resolution. One of the greatest challenges posed in applications of EEG is the unstable signal quality susceptible to inevitable artifacts during recordings. To date, most existing techniques for EEG artifact removal and reconstruction are applicable to offline analysis solely, or require individualized training data to facilitate online reconstruction. We have proposed CLEEGN, a novel convolutional neural network for plug-and-play automatic EEG reconstruction. CLEEGN is based on a subject-independent pre-trained model using existing data and can operate on a new user without any further calibration. The performance of CLEEGN was validated using multiple evaluations including waveform observation, reconstruction error assessment, and decoding accuracy on well-studied labeled datasets. The results of simulated online validation suggest that, even without any calibration, CLEEGN can largely preserve inherent brain activity and outperforms leading online/offline artifact removal methods in the decoding accuracy of reconstructed EEG data. In addition, visualization of model parameters and latent features exhibit the model behavior and reveal explainable insights related to existing knowledge of neuroscience. We foresee pervasive applications of CLEEGN in prospective works of online plug-and-play EEG decoding and analysis.

Figures (16)

• Figure 1: The denoising flow diagram with the proposed artifact removal architecture. $X$, $Y$, $N$ denote the observed EEG data, the ideal EEG signals and the noise respectively. The reference EEG data, $\tilde{Y}_r$, is the noiseless estimation generated by existing artifact removal method in the offline scenario.
• Figure 2: The waveform of the simulated sources
• Figure 3: An example of the synthesized signal mixture (black) and the corresponding reference signals (red)
• Figure 4: An experimental result waveform for eliminating noise sources from an synthetic signal under scenario-2. The first raw illustrates the raw signal (black) without suppressing noise and the ground truth signal (red). The following from top to bottom, the waveform colored in blue are denoised by our method, IC-U-Net, 1D-ResCNN, SCNN and RNN.
• Figure 5: Reconstruction of synthetic data in power spectral density (PSD) using various methods under scenario-2. The top-left subfigure compares the PSD of the raw signal mixture (gray) with the reference signal (red). The remaining subfigures compare the reconstruction signals (blue) from different methods to the reference signals (red).