Detection of P300 waves in single trials by the wavelet transform (WT)

Area of Science:

• Cognitive Neuroscience
• Electrophysiology
• Signal Processing

Background:

• The P300 event-related potential (ERP) is typically averaged from task-relevant stimuli in oddball paradigms.
• Cognitive state fluctuations during experiments cause variability in ERP components, potentially obscuring P300 detection with conventional methods.

Purpose of the Study:

• To develop and validate a response-based classification procedure for identifying P300 components in single electroencephalography (EEG) sweeps.
• To investigate the influence of cognitive state changes on P300 presence, independent of task performance.

Main Methods:

• Utilized a response-based classification using wavelet transform (WT) on single EEG sweeps from an auditory oddball paradigm.
• Identified P300 by the positivity of the 4th delta (0.5-4 Hz) wavelet coefficient (310-430 ms post-stimulus).
• Compared averaged P300 waves from response-selected sweeps against conventional target and non-target averages.

Main Results:

• Response-selected sweep averaging yielded enhanced P300 waves compared to conventional target averaging.
• Averaging non-target sweeps showed significantly less positivity in the P300 latency range.
• Combining task-based and response-based criteria revealed significant differences, indicating cognitive state impact.

Conclusions:

• The wavelet transform-based response classification effectively identifies P300 components in single EEG trials.
• Cognitive state variations significantly influence P300 presence, a factor not fully captured by task relevance alone.
• This method offers improved P300 detection and a more nuanced understanding of cognitive processing during EEG experiments.