Accuracy of two dipolar inverse algorithms applying reciprocity for forward calculation

Area of Science:

• Neuroscience
• Biomedical Engineering
• Computational Biology

Background:

• The electroencephalography (EEG) inverse problem is crucial for localizing neural activity.
• Accurate source localization requires efficient forward computation and robust inverse algorithms.
• Previous methods often face computational challenges and sensitivity to noise.

Purpose of the Study:

• To evaluate two inverse algorithms for EEG source localization using a single dipole model.
• To assess the efficiency and accuracy of the lead field approach in forward computations.
• To compare the performance of least-squares and probability-based methods under varying noise conditions.

Main Methods:

• Applied two inverse algorithms (least-squares and probability-based) for EEG source localization.
• Utilized the lead field approach based on the reciprocity theorem for efficient forward computations.
• Employed a realistic five-compartment volume conductor model and simulated dipole sources with and without noise.

Main Results:

• Dipole localization errors ranged from 0-9 mm without noise and 2-22 mm with noise.
• Both inverse algorithms demonstrated similar performance.
• The lead field method proved effective for solving the EEG inverse problem, especially for multiple sources or time instances.

Conclusions:

• The lead field method is a viable and efficient approach for EEG inverse problem solving.
• Accurate source localization is achievable even with noisy EEG data.
• This methodology is particularly beneficial for analyzing complex neural activity patterns over time.