Comparative Study of ICA Results under Different Conditional ERP Datasets

JC Zhang,L Yao,XJ Zhao

DOI: https://doi.org/10.1109/icnnb.2005.1614935

2005-01-01

Abstract:The same subject under different condition (experimental events) will induce different brain currents spreading through his/her head, and produce different event-related potential (ERP) datasets from different experimental events. Comparison of these different ERP datasets is an effective method to study the brain dynamics. But ERP waveforms recorded at scalp sensors are a sum of temporally independent and spatially fixed components arising from distinct or overlapping brain or extra-brain sources. In order to compare the original source of different ERP datasets, we propose a new approach to find the common or similar ICA components in these two ERP datasets. Here two related ERP datasets are blindly separated into two sets of temporally independent and spatially fixed components by independent component analysis (ICA), and then Pearson correlation coefficients were calculated for ICA components and ICA projection vectors between these two datasets. Based on these Pearson correlation coefficients we can find some similar or common components in these different ERP datasets. The results also show that the approach can identify those similar or common components, which also proves an effective and efficient method to study the brain dynamics