P300 component identification using source analysis techniques: reduced latency variability.

ABSTRACT

P300 latency variability in normal subjects is a complicating factor in clinical event-related potential studies because it limits diagnostic applicability. The current study was conducted to determine whether identification of P300 (P3A and P3B) components using source analysis techniques can reduce variability in P300 parameters. Data were recorded with a 128-channel EEG system in 18 healthy subjects. The authors used a standard, auditory two-tone oddball paradigm with targets of 2000 Hz and standards of 1000 Hz. Two simple source analysis models with one or two rotating dipoles were applied to grand average data and individual data. Dipole time courses were combined with mapping results to extract P3A and P3B component latencies. Latencies obtained with conventional P300 analysis were compared with source analysis results. The source analysis method identified both P3A and P3B components in a substantially larger percentage of subjects (88% vs. 33%) than the conventional method. The source analysis method yielded a later mean P3B latency (357 msec vs. 323 msec, P < 0,001) with a smaller standard deviation (9 msec vs. 23 msec, P = 0,003) than the conventional P300 method. The relative contribution of the temporally separate P3A and P3B components to the P300 complex amplitude is highly variable. This explains the larger latency standard deviation in conventional P300 analysis. The source analysis method was able to identify P300 components in a large percentage of the cases. The result is a considerable reduction of P300 latency variability in normal subjects. This could have important consequences for clinical event-related potential research, because diagnostic sensitivity and specificity of P300 latency may improve with this method.