A systematic evaluation of the spherical model accuracy in EEG dipole localization

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Accurate source localization of brain activity using electroencephalography (EEG) is crucial for understanding neural processes.
• Realistic head models are complex, leading to the common use of simplified spherical head models in EEG inverse problems.
• The intrinsic bias introduced by these simplifications is not fully understood.

Purpose of the Study:

• To quantify the localization error bias caused by using spherical geometry models for EEG data.
• To assess the impact of electrode number on this bias.
• To evaluate the effect of multiple simultaneous sources on localization accuracy.

Main Methods:

• Simulated EEG data generated from a realistically shaped head model with 2000 cortical dipoles.
• Forward and inverse calculations using four different spherical models and one uniformly meshed model.
• Analysis of localization errors based on dipole depth and location.
• Investigation of the effect of varying electrode configurations (19, 32, and 63 electrodes).
• Simulations with two simultaneously active dipoles.

Main Results:

• The best spherical model resulted in localization errors of 5-6 mm in the upper head and 15-25 mm in the lower head.
• Increasing electrodes from 19 to 32 improved localization by an average of 2.7 mm.
• Further increasing electrodes from 32 to 63 yielded improvements of less than 1 mm.
• Simultaneous dipole activity increased localization errors by approximately 2-3 mm.

Conclusions:

• Spherical head models introduce substantial intrinsic localization bias in EEG, particularly for deeper sources.
• While increasing electrode density can mitigate some error, the benefits diminish significantly beyond 32 electrodes.
• These findings highlight the limitations of simplified models and inform the development of more accurate EEG source localization techniques.