文献类型：期刊文献

英文题名：Research progress on emotion recognition based on electroencephalogram signals

作者：Li, Xue[1,2];Yan, Zuojian[2];Gong, Piqiang[1,2];Lin, Dongmei[3];Chen, Fuming[2]

第一作者：薛莉;Li, Xue

通信作者：Chen, FM[1]

机构：[1]Gansu Univ Chinese Med, Sch Med Informat Engn, Dept Biomed Engn, Lanzhou, Peoples R China;[2]940th Hosp Joint Logist Support Force Chinese Peop, Med Secur Ctr, Lanzhou 730050, Gansu, Peoples R China;[3]Lanzhou Univ Technol, Coll Elect & Informat Engn, Lanzhou, Peoples R China

第一机构：甘肃中医药大学

通信机构：[1]corresponding author), 940th Hosp Joint Logist Support Force Chinese Peop, Med Secur Ctr, Lanzhou 730050, Gansu, Peoples R China.

年份：2026

卷号：113

外文期刊名：BIOMEDICAL SIGNAL PROCESSING AND CONTROL

收录：;WOS:【SCI-EXPANDED(收录号:WOS:001631094100001)】；

基金：This work was funded by Natural Science Foundation of China (61901515, 62361038) Natural Science Foundation of Gansu Province (22JR5RA002) and the Innovation Fund for University Teachers of Department of Education of Gansu Province (2023A-019) .

语种：英文

外文关键词：Emotion recognition; Affective computing; Electroencephalogram signals; Deep learning

摘要：Emotion recognition, a key subfield of affective computing (AC), has garnered significant attention in recent years due to its promising applications and its broad interdisciplinary impact. Electroencephalogram (EEG), originating directly from the brain's cortical activity and being resistant to artifacts, is a dependable and objective indicator of emotional states. The swift advancements in Artificial Intelligence (AI), especially within the realm of deep learning (DL), have significantly propelled the development of this field. DL algorithms, including convolutional neural networks (CNN), capsule networks (CapsNet), recurrent neural networks (RNN), and hybrid networks, have found widespread application in EEG-based emotion recognition, which significantly improve the accuracy of feature extraction, spatio-temporal pattern recognition and classification, and effectively deal with the complexity of emotion states. In addition, EEG emotion recognition has important interdisciplinary potential, which has driven progress across multiple disciplines, including computer science, psychology, neuroscience, and medicine. It also showcases significant potential in advancing both research and clinical assessment of neurological conditions, such as sleep disorders, schizophrenia, epilepsy, and multiple sclerosis. Given the relatively small amount of literature reviewed in this field, this paper provides a comprehensive overview of the fundamental theories behind emotion modeling, the five EEG signal bands, the impact of various brain regions on emotional states, the commonly used evaluation indexes, and the methodology for recognizing emotions through EEG signals, and analyzes the application of DL algorithms in emotion recognition, which is aimed at helping researchers to quickly master the core knowledge in this field and providing a reference for subsequent research.