Systematic source estimation of spikes by a combination of independent component analysis and RAP-MUSIC

Efficient unsupervised algorithms for the detection of seizures in continuous EEG recordings from rats after brain injury

“…Subsequent improvements included detection of the state of the patient (sleep versus awake) and elimination of artifacts associated with that state Wang, 1991, 1992). Other algorithms have been employed using such techniques as neural networks (Gabor et al, 1996;Gabor, 1998;James et al, 1999;Ko and Chung, 2000), independent component analysis (Kobayashi et al, 1999(Kobayashi et al, , 2001(Kobayashi et al, , 2002, and wavelet analysis (Khan and Gotman, 2003). Most recently, non-linear techniques (Jing and Takigawa, 2000;Lehnertz, 1999;Li et al, 2003;Litt and Echauz, 2002;Navarro et al, 2002) have been employed using such quantities as the correlation dimension, Lyapunov exponent, Kolmogrov entropy, marginal predictability and similarity index.…”

A novel method for automated classification of epileptiform activity in the human electroencephalogram-based on independent component analysis

“…PCA and ICA are ways of extracting empirical bases that differ in subtle, but important, respects [17]. Indeed, ICA has previously been shown to be useful for analyzing epileptiform discharges, but only given a degree of hand-tuning [26][27][28]32]. We employed a novel variant of ICA in a fully automatic procedure, and showed that it out-performs PCA and is at least competitive with Reveal on the same input dataset.…”

Combining the extremities on the basis of separation: a new approach to EEG/ERP source localization

“…To date, many simulation and real data studies demonstrated the usefulness of Principal Component Analysis (PCA) followed by rotation, BSS based on Second Order Statistics (such as implemented in SOBI algorithm), and Independent Component Analysis (ICA) for improving source localization [4][5][6][7][8][9][10][11][12][13]. All these BSS techniques, theoretically, may help to localize even very weak and strongly overlapped sources, and to locate much higher number of dipole-like sources than can be localized without such preprocessing because the number of the dipoles which can be fitted is limited, in the case that separation is performed perfectly, only by the number of separated sources, which is, in the case of ICA and many other BSS techniques, usually equal to the number of electrodes.…”