Proc. of the Australian and New Zealand Intelligent Information Systems Conference (ANZIIS), pp. 361-366, Sydney, 2003
Detection of brain activity for averaged multiple-trial magnetoencephalography data
Y. Konno, J. Cao, T. Arai and T. Takeda
Abstract: Treating averaged multiple-trial data is a common approach in recent papers for applying independent component analysis (ICA) to neurobiological signal prodessing. Although averaging increases the signal-to-noise ratio (SNR), important information is lost, such as the strength of an evoked response and its dynamics. Alteratively, when averaging few-trial data, not much information is lost, but SNR is very poor. In this paper, we deal with averaged data affew as well as many trials, and we demonstrate that not only the location, but also the direction vector and dipole moment of evoked fields (EFs) can be obtained by applying our method, even when the mumber of trials is small.