IC-U-Net: A U-Net-based Denoising Autoencoder Using Mixtures of Independent Components for Automatic EEG Artifact Removal

作     者：Chuang, Chun-Hsiang Chang, Kong-Yi Huang, Chih-Sheng Jung, Tzyy-Ping 

作者机构：Natl Tsing Hua Univ Coll Educ Res Ctr Educ & Mind Sci Hsinchu Taiwan Natl Tsing Hua Univ Inst Informat Syst & Applicat Coll Elect Engn & Comp Sci Hsinchu Taiwan Natl Tsing Hua Univ Dept Educ & Learning Technol Hsinchu Taiwan Natl Taiwan Ocean Univ Dept Comp Sci & Engn Keelung Taiwan Elan Microelect Corp Dept Artificial Intelligence Res & Dev Hsinchu Taiwan Natl Yang Ming Chiao Tung Univ Coll Artificial Intelligence & Green Energy Hsinchu Taiwan Natl Taipei Univ Technol Coll Elect Engn & Comp Sci Taipei Taiwan Univ Calif San Diego Inst Engn Med La Jolla CA USA Univ Calif San Diego Inst Neural Computat La Jolla CA USA 

出 版 物：《NEUROIMAGE》 (神经图像)

年 卷 期：2022年第263卷

页      面：119586-119586页

核心收录：

学科分类：1002[医学-临床医学] 1001[医学-基础医学(可授医学、理学学位)] 1010[医学-医学技术（可授医学、理学学位）] 1009[医学-特种医学] 10[医学] 

基　　金：Ministry of Science and Technology, Taiwan [MOST 111-2636-E-007-020, 110-2636-E-007-018, 109-2636-E-007-022] Research Center for Education and Mind Sciences, National Tsing Hua University Yin Shu-Tien Educational Foundation [111F7120A8] 

主　　题：EEG Artifact Removal Signal Reconstruction U-Net Independent Component Analysis ICLabel Denoising Autoencoder Deep Learning 

摘      要：Electroencephalography (EEG) signals are often contaminated with artifacts. It is imperative to develop a practical and reliable artifact removal method to prevent the misinterpretation of neural signals and the underperformance of brain-computer interfaces. Based on the U-Net architecture, we developed a new artifact removal model, IC-U-Net, for removing pervasive EEG artifacts and reconstructing brain signals. IC-U-Net was trained using mixtures of brain and non-brain components decomposed by independent component analysis. It uses an ensemble of loss functions to model complex signal fluctuations in EEG recordings. The effectiveness of the proposed method in recovering brain activities and removing various artifacts (e.g., eye blinks/movements, muscle activities, and line/channel noise) was demonstrated in a simulation study and four real-world EEG experiments. IC-U-Net can reconstruct a multi-channel EEG signal and is applicable to most artifact types, offering a promising end -to-end solution for automatically removing artifacts from EEG recordings. It also meets the increasing need to image natural brain dynamics in a mobile setting. The code and pre-trained IC-U-Net model are available at https://***/roseDwayane/AIEEG .