SEEG dipole source localization based on an empirical Bayesian approach taking into account forward model...

Area of Science:

• Neuroscience
• Biophysics
• Computational Biology

Background:

• Electromagnetic brain source localization is an ill-posed inverse problem requiring regularization.
• Sparsity constraints are increasingly used, assuming focal brain activity during specific events.
• The accuracy of the forward model, often based on approximations, is critical but rarely questioned.

Purpose of the Study:

• To develop an empirical Bayesian approach to address uncertainties in the forward model for brain source localization.
• To improve the accuracy and sparsity of source localization using SEEG (stereoelectroencephalography) measurements.

Main Methods:

• An empirical Bayesian framework was employed to incorporate constrained variations of a prior physical model.
• The method specifically addresses uncertainties in the forward model construction for SEEG data.
• Simulations and real SEEG signals were used to validate the approach.

Main Results:

• The proposed method enhanced the accuracy of source time-course estimation.
• Improved sparsity of the resulting source activation maps was observed.
• The approach demonstrated applicability to real-world SEEG data.

Conclusions:

• Accounting for forward model uncertainties within an empirical Bayesian framework improves brain source localization accuracy and sparsity.
• This method offers a robust solution for SEEG data analysis.
• The findings highlight the importance of addressing forward model uncertainties in neuroimaging.