ART: Artifact Removal Transformer for Reconstructing Noise-Free Multichannel Electroencephalographic Signals

Chun-Hsiang Chuang,Kong-Yi Chang,Chih-Sheng Huang,Anne-Mei Bessas

2024-09-11

Abstract:Artifact removal in electroencephalography (EEG) is a longstanding challenge that significantly impacts neuroscientific analysis and brain-computer interface (BCI) performance. Tackling this problem demands advanced algorithms, extensive noisy-clean training data, and thorough evaluation strategies. This study presents the Artifact Removal Transformer (ART), an innovative EEG denoising model employing transformer architecture to adeptly capture the transient millisecond-scale dynamics characteristic of EEG signals. Our approach offers a holistic, end-to-end denoising solution for diverse artifact types in multichannel EEG data. We enhanced the generation of noisy-clean EEG data pairs using an independent component analysis, thus fortifying the training scenarios critical for effective supervised learning. We performed comprehensive validations using a wide range of open datasets from various BCI applications, employing metrics like mean squared error and signal-to-noise ratio, as well as sophisticated techniques such as source localization and EEG component classification. Our evaluations confirm that ART surpasses other deep-learning-based artifact removal methods, setting a new benchmark in EEG signal processing. This advancement not only boosts the accuracy and reliability of artifact removal but also promises to catalyze further innovations in the field, facilitating the study of brain dynamics in naturalistic environments.

Signal Processing,Machine Learning

What problem does this paper attempt to address?

The paper aims to address the issue of artifact removal in electroencephalogram (EEG) signals. Specifically, the paper proposes a new model called the Artifact Removal Transformer (ART) for efficiently removing various artifact noises from multi-channel EEG data. This challenge has long affected neuroscience research and brain-computer interface (BCI) performance. To tackle this challenge, the research team adopted the Transformer architecture to capture the unique millisecond-level dynamic characteristics of EEG signals and proposed an end-to-end denoising solution.

Additionally, the paper generated a large number of noise-clean EEG data pairs through Independent Component Analysis (ICA), thereby enhancing the effectiveness of supervised learning. The researchers conducted extensive validation using multiple public datasets, including datasets from different BCI applications, and evaluated the performance using metrics such as Mean Squared Error (MSE) and Signal-to-Noise Ratio (SNR). The results show that ART outperforms other deep learning-based artifact removal methods in EEG signal processing, setting a new benchmark for EEG signal processing.

This study not only improves the accuracy and reliability of artifact removal but also promotes further innovation in the study of brain dynamics in natural environments.