TS-ADMM: A Novel EEG Source Localization Method Based on Time Smoothing Optimization

The human brain is a complex system composed of billions of neurons. Accurately estimating the location of brain activity is an important step in understanding how the brain works. Due to the rich temporal information in electroencephalography (EEG) signals, many studies have proved that utilizing the temporal information in EEG signals can improve the performance of EEG source localization. Alternating direction method of multipliers (ADMM) is a commonly used method for EEG source localization, but it is greatly affected by spurious signals. In this paper, we propose a novel method called TS-ADMM. First, we utilize the fast iterative shrinkage-thresholding (FISTA) method to obtain the initial value of the brain cortical potential matrix. Then, we optimize the time smoothing of the graph matrix to improve ADMM, which effectively reduces the impact of spurious signals on localization. Numerical experiments show that compared with ADMM, TS-ADMM can reduce the EEG source localization error by 15.5%.