The independent components of auditory P300 and CNV evoked potentials derived from single-trial recordings

Abstract

The back-projected independent components (BICs) of single-trial, auditory P300 and contingent negative variation (CNV) evoked potentials (EPs) were derived using independent component analysis (ICA) and cluster analysis. The method was tested in simulation including a study of the electric dipole equivalents of the signal sources. P300 data were obtained from healthy and Alzheimer's disease (AD) subjects. The BICs were of approximately 100 ms duration and approximated positive- and negative-going half-sinusoids. Some positively and negatively peaking BICs constituting the P300 coincided with known peaks in the averaged P300. However, there were trial-to-trial differences in their occurrences, particularly where a positive or a negative BIC could occur with the same latency in different trials, a fact which would be obscured by averaging them. These variations resulted in marked differences in the shapes of the reconstructed, artefact-free, single-trial P300s. The latencies of the BIC associated with the P3b peak differed between healthy and AD subjects (p < 0.01). More reliable evidence than that obtainable from single-trial or averaged P300s is likely to be found by studying the properties of the BICs over a number of trials. For the CNV, BICs corresponding to both the orienting and the expectancy components were found.

Waveform Analysis of Non-Oscillatory Independent Components in Single-Trial Auditory Event-Related Activity in Healthy Subjects and Alzheimers Disease Patients

“…Other authors based their conclusions on the activations [7,25,26,28,[30][31][32], which are indeterminate in magnitude and sign. While this may be useful, we prefer to study the BICs, which are correct in magnitude and sign [12], since direct comparisons can be made between their topographies from trial-to-trial, and with the established topographies of averaged ERPs. Further, the BICs represent the components of the measured electrode signals, and, using LORETA [9], [10] the locations of the electrical current dipoles representing their sources may be calculated.…”

“…Further, the BICs represent the components of the measured electrode signals, and, using LORETA [9], [10] the locations of the electrical current dipoles representing their sources may be calculated. We based our analysis of the individual P300 waveform on a single trial methodology [12], in which averaging and concatenation were avoided, in order to characterise the BICs and seek biomarkers for AD.…”

“…Jervis et al [12,34,35] described the variability of occurrence in the polarities, latencies, and topographies of the observed BICs and their time-variations. These behaviours were used to explain some of the characteristics of the averaged P300 waveform.…”

“…The individual source signals may be computed from the measured scalp voltages using ICA [5,[37][38][39]. In [12] it was explained that if S be a matrix of temporally independent source signals and Y be the matrix of measured signals at the electrodes (EEGs i y ), which is assumed to consist of linear sums of the source signals ( S ) which have passed through an unknown, linear transmission system characterised by an m x m mixing matrix, A , then we may write…”

“…In this paper an extended and formalised study of earlier work [12] on the short-duration, non-oscillatory, halfsinusoidal BICs constituting the single trial auditory ERP waveforms measured in healthy subjects and Alzheimer's Disease patients is presented. More specifically, we have used the topographies of the BICs in the secondary clustering obtained via source localization methods instead of the numerically inexact coefficient maps.…”