A Novel Multi-Stage Algorithm for Real-Time Detection and Correction of Ocular Artifacts in EEG: A Calibration-Free...

Area of Science:

• Neuroscience and Biomedical Engineering
• Signal Processing

Background:

• Ocular artifacts, such as blinks, significantly corrupt electroencephalographic (EEG) signals.
• Existing artifact correction methods often require calibration or additional sensors, limiting real-world applicability.

Purpose of the Study:

• To introduce CFo-CLEAN, a novel online method for detecting and correcting ocular artifacts from EEG signals without prior calibration.
• To evaluate the performance of CFo-CLEAN against established artifact correction techniques.

Main Methods:

• The CFo-CLEAN method utilizes an Enhanced Adaptive Data-driven Algorithm (eADA) for dynamic, real-time artifact identification and correction directly from EEG data.
• EEG data from 38 participants during real-world driving scenarios were used for evaluation.
• Performance was compared against Independent Component Analysis (ICA), regression, and subspace reconstruction methods.

Main Results:

• CFo-CLEAN effectively reduced ocular artifact contamination in EEG signals.
• The method demonstrated good preservation of neurophysiological content.
• Longer time windows (90 seconds) in CFo-CLEAN implementations showed superior EEG signal preservation, especially in higher frequency bands.

Conclusions:

• CFo-CLEAN is a viable and effective method for real-time ocular artifact correction in EEG.
• Its adaptive nature and lack of calibration requirement make it suitable for mobile and dynamic environments.
• The method advances the development of brain-computer interfaces (BCIs), neuroergonomics, and cognitive monitoring systems.