文献类型：期刊文献

英文题名：A Growing Bubble Speller Paradigm for Brain-Computer Interface Based on Event-Related Potentials

作者：Jin, Jing[1];Zhao, Xueqing[1];Daly, Ian[3];Li, Shurui[2];Wang, Xingyu[1];Cichocki, Andrzej[4,5];Jung, Tzyy-Ping[6,7]

机构：[1]East China Univ Sci & Technol, Minist Educ, Key Lab Smart Mfg Energy Chem Proc, Shanghai 200237, Peoples R China;[2]East China Univ Sci & Technol, Sch Math, Shanghai, Peoples R China;[3]Univ Essex, Sch Comp Sci & Elect Engn, Brain Comp Interfacing & Neural Engn Lab, Colchester, England;[4]Polish Acad Sci, Syst Res Inst, Warsaw, Poland;[5]Nicolaus Copernicus Univ UMK, Torun, Poland;[6]Univ Calif San Diego, Inst Neural Computat, Swartz Ctr Computat Neurosci, La Jolla, CA USA;[7]Univ Calif San Diego, Inst Engn Med, Ctr Adv Neurol Engn, La Jolla, CA USA

年份：2025

卷号：72

期号：3

起止页码：1188

外文期刊名：IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING

收录：;EI(收录号：20244617358648);WOS:【SCI-EXPANDED(收录号:WOS:001440187500025)】；

基金：This work was supported in part by the National Natural Science Foundation of China under Grant 62176090 and in part by STI 2030-major projects under Grant 2022ZD0208900, in part by the Shanghai Municipal Science and Technology Major Projectunder Grant 2021SHZDZX. This research is also supported in part by Project of Jiangsu Province Science and Technology Plan Special Fundin 2022, in part by Key Research and Development Plan Industry Fore-sight, Fundamental Research Fund for the Central Universities under Grant JKH01231636 and in part by Key Core Technologies under Grant BE2022064-1.

语种：英文

外文关键词：Image color analysis; Electroencephalography; Visualization; Feature extraction; Decoding; Faces; Electric potential; Brain-computer interfaces; Support vector machines; Convolutional neural networks; Brain-computer interface (BCI); event-related potential (ERP); speller paradigm; growing bubble; multiple windows

摘要：Objective: Event-related potentials (ERPs) reflect electropotential changes within specific cortical regions in response to specific events or stimuli during cognitive processes. The P300 speller is an important application of ERP-based brain-computer interfaces (BCIs), offering potential assistance to individuals with severe motor disabilities by decoding their electroencephalography (EEG) to communicate. Methods: This study introduced a novel speller paradigm using a dynamically growing bubble (GB) visualization as the stimulus, departing from the conventional flash stimulus (TF). Additionally, we proposed a "Lock a Target by Two Flashes" (LT2F) method to offer more versatile stimulus flash rules, complementing the row and column (RC) and single character (SC) modes. We applied the "Sub and Global" multi-window mode to EEGNet (mwEEGNet) to enhance classification and explored the performance of eight other representative algorithms. Results: Twenty healthy volunteers participated in the experiments. Our analysis revealed that our proposed pattern elicited more pronounced negative peaks in the parietal and occipital brain regions between 200 ms and 230 ms post-stimulus onset compared with the TF pattern. Compared to the TF pattern, the GB pattern yielded a 2.00% increase in online character accuracy (ACC) and a 5.39 bits/min improvement in information transfer rate (ITR) when using mwEEGNet. Furthermore, results demonstrated that mwEEGNet outperformed other methods in classification performance. Conclusion and Significance: These results underscore the significance of our work in advancing ERP-based BCIs.