Authors:
F. Grosselin
1
;
Y. Attal
2
and
M. Chavez
3
Affiliations:
1
Sorbonne Universités, UPMC Univ. Paris 06, Inserm U-1127, CNRS UMR-7225, Institut du Cerveau et de la Moelle Épinière (ICM), Groupe Hospitalier Pitié Salpêtrière-Charles Foix, 75013, Paris, France, myBrainTechnologies, 75010, Paris and France
;
2
myBrainTechnologies, 75010, Paris and France
;
3
CNRS UMR-7225, Groupe Hospitalier Pitié Salpêtrière-Charles Foix, 75013, Paris and France
Keyword(s):
EEG, Alpha Peak Frequency, IAF Estimation.
Related
Ontology
Subjects/Areas/Topics:
Biomedical Engineering
;
Biomedical Instruments and Devices
;
Brain-Computer Interfaces
;
Devices
;
EMG Signal Processing and Applications
;
Health Engineering and Technology Applications
;
Human-Computer Interaction
;
Neural Signal Processing
;
NeuroSensing and Diagnosis
;
Neurotechnology, Electronics and Informatics
;
Physiological Computing Systems
Abstract:
We present a method to determine the individual alpha (α) peak frequency (IAF) of EEG segments. The algorithm uses information over previous time-windows to determine the current IAF. First, the 1/ f trend of the spectrum is estimated by an iterative curve-fitting procedure and then removed from the spectrum. Finally, local maxima are identified in the corrected spectrum. If an α peak is ambiguous, i.e. when several peaks are observed due to different physiological α activations or to a noisy spectral activity, the algorithm selects the most probable one based on the peaks detected in previous time windows. This approach allows the detection of small α activities and ensures a precise and stable detection of the α peak, without offline analysis or a prior estimation of a reference spectrum. This is particularly important for real-time applications like α-based neurofeedback for which a precise and stable feedback is required for an efficient learning.