Keep it simple: a case for using classical minimum norm estimation in the analysis of EEG and MEG data.

Area of Science:

• Bioelectromagnetism
• Computational Neuroscience
• Medical Imaging

Background:

• Bioelectromagnetic inverse problems require source localization from limited sensor data.
• A priori information about source characteristics is often unavailable or unreliable.
• Existing inverse solution methods vary in their handling of missing information.

Purpose of the Study:

• To determine the optimal inverse solution for bioelectromagnetic inverse problems when no reliable a priori information is available.
• To compare maximum-likelihood, minimum norm, and resolution optimization approaches theoretically.
• To identify the fundamental limitations of electroencephalography (EEG) and magnetoencephalography (MEG) in source localization.

Main Methods:

• Theoretical comparison of three inverse solution approaches: maximum-likelihood, minimum norm, and resolution optimization.
• Analysis of how a priori information influences solutions within these frameworks.
• Identification of the minimum norm pseudoinverse (MNP) as the solution in the absence of a priori information.

Main Results:

• All three compared frameworks yield identical solutions when the same a priori information is applied.
• The minimum norm pseudoinverse (MNP) is the common solution when no a priori information is available.
• Limitations in depth localization are inherent to recording modalities, not specific to the MNP method.

Conclusions:

• The minimum norm solution reflects the information content inherent in the data, making it suitable for assessing EEG/MEG resolution and accuracy limits.
• The minimum norm solution is a valuable approach for bioelectromagnetic inverse problems lacking reliable a priori source information.
• This method is particularly useful for analyzing complex cognitive tasks and noisy data, such as single-trial EEG/MEG recordings.