Motion and Muscle Artifact Removal Validation Using an Electrical Head Phantom, Robotic Motion Platform, and Dual...

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Motion and muscle artifacts significantly degrade electroencephalography (EEG) signal quality during locomotion.
• Accurate brain signal recovery is crucial for understanding neural activity during movement.

Purpose of the Study:

• To evaluate methods for recovering ground-truth artificial brain signals from artifact-contaminated EEG recordings.
• To determine optimal strategies for removing motion and muscle artifacts using advanced signal processing techniques.

Main Methods:

• Utilized an electrical head phantom with simulated brain and muscle sources, subjected to robotic motion.
• Recorded 128-channel dual-layer EEG and 8-channel neck electromyography (EMG).
• Applied Independent Component Analysis (ICA), Artifact Subspace Reconstruction (ASR), and Canonical Correlation Analysis (CCA) to artifact-contaminated data.

Main Results:

• Combining isolated noise and EMG recordings in ICA decomposition significantly improved brain signal recovery.
• A reduced set of 32 noise and 6 EMG channels yielded performance comparable to full arrays.
• ASR and CCA preprocessing enhanced source separation, with ASR's effectiveness dependent on muscle activity amplitude.
• Integrated artifact removal strategies improved overall source signal recovery.

Conclusions:

• Integrating noise and EMG recordings into ICA, alongside ASR and CCA, is an effective strategy for artifact removal in EEG during locomotion.
• Optimized channel subsets can maintain high performance, reducing computational load.
• This study provides a robust framework for enhancing EEG signal integrity in ambulatory research.