Finding tau rhythms in EEG: An independent component analysis approach

Area of Science:

• Neuroscience
• Auditory Neuroscience
• Signal Processing

Background:

• Tau rhythms are sound-responsive alpha band oscillations (~8-13 Hz) originating in auditory brain regions.
• Detecting tau rhythms in electroencephalogram (EEG) is challenging, often requiring magnetoencephalography or intracranial recordings.

Purpose of the Study:

• To demonstrate the effectiveness of independent component analysis (ICA) for identifying and studying tau rhythms in EEG data.
• To optimize EEG analysis parameters for reliable tau rhythm detection.

Main Methods:

• EEG data from 18 subjects exposed to acoustic stimuli were analyzed using various high-pass and low-pass filters.
• Four ICA algorithms (fastICA, infomax, AMICA, mAMICA) were applied to identify tau-related independent components (ICs).
• Tau ICs were characterized by their localization near the superior temporal gyri and alpha band spectral peaks.

Main Results:

• ICA successfully identified tau-related ICs in ~94% of subjects when using optimal parameters (aggressive high-pass filters and mAMICA).
• Tau ICs exhibited alpha suppression during sound presentation, distinct from other alpha-band ICs.
• The choice of high-pass filter cutoff and ICA algorithm significantly influenced the detection rate of tau ICs, with mAMICA performing best.

Conclusions:

• ICA decomposition, particularly with aggressive high-pass filtering and the mAMICA algorithm, is a viable method for detecting tau rhythms in EEG.
• This approach enhances the ability to study auditory-responsive tau rhythms, potentially increasing research discoveries in this area.
• The findings suggest EEG-based tau rhythm analysis using ICA can yield results comparable to MEG and intracranial recordings.