Consequences of EEG electrode position error on ultimate beamformer source reconstruction performance.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Inaccurate electroencephalography (EEG) electrode coordinates introduce errors in forward models and source reconstruction.
• Errors stem from digitization noise and manual coregistration with MRI.
• Previous assumptions suggested errors primarily cause source displacement, underestimating their impact.

Purpose of the Study:

• To quantify the impact of EEG electrode registration errors on source reconstruction performance.
• To compare the accuracy of different electrode digitization techniques.
• To evaluate the effect of registration accuracy on beamformer output Signal-to-Noise Ratio (SNR).

Main Methods:

• Measured electrode positions on a 3D-printed head replica using fringe projection, Flying Triangulation, and electromagnetic digitization.
• Used fringe projection data as ground truth for accuracy assessment.
• Simulated EEG data with varying electrode registration errors and sensor noise levels.
• Processed simulated data using a beamformer to assess performance.

Main Results:

• Flying Triangulation sensor achieved a mean error of 1.5 mm, significantly better than the electromagnetic digitizer's 6.8 mm mean error.
• Beamformer output SNR was severely degraded with electromagnetic digitizer coordinates compared to Flying Triangulation.
• SNR penalties reached several decibels, particularly with lower-accuracy digitization.

Conclusions:

• Electrode registration accuracy critically affects beamformer performance.
• Substantial degradation in output SNR can occur due to seemingly minor errors.
• Accurate sensor coregistration is essential for resolving low-SNR signals from deeper brain structures or gamma-band activity.