Integrated approach of an artificial neural network and numerical analysis to multiple equivalent current dipole...

Area of Science:

• Neuroscience
• Biomedical Engineering
• Computational Neuroscience

Background:

• Electroencephalography (EEG) is crucial for studying brain activity.
• Accurate localization of neural sources is challenging.
• Existing methods for equivalent current dipole (ECD) localization have limitations.

Purpose of the Study:

• To develop a PC-based multichannel EEG measurement and analysis system.
• To implement a novel neural network and numerical analysis (NNN) approach for rapid and precise ECD localization.
• To integrate EEG source localization with magnetic resonance imaging (MRI).

Main Methods:

• Development of a 64-channel EEG recording system.
• Measurement of electrode positions in 3D space.
• Application of a hybrid approach combining a feedforward artificial neural network (ANN) for initialization and Powell's hybrid method for refinement of ECD parameters.
• Superimposition of localized ECDs onto MRI.

Main Results:

• The system allows simultaneous recording of up to 64 EEG channels.
• The NNN approach provides high-speed and noise-robust ECD localization.
• Accurate refinement of ECD parameters was achieved.
• The method was successfully applied to sensory evoked potentials and event-related potentials.

Conclusions:

• The developed PC-based EEG system offers advanced capabilities for neural activity measurement and analysis.
• The novel ANN-based ECD localization method significantly improves speed and accuracy.
• This integrated system facilitates more precise understanding of brain function through advanced EEG source localization.