Imaging brain activation streams from optical flow computation on 2-Riemannian manifolds

Area of Science:

• Neuroscience
• Computer Vision
• Applied Mathematics

Background:

• Magnetoencephalography (MEG) and Electroencephalography (EEG) provide high temporal resolution insights into brain activity.
• Analyzing spatiotemporal dynamics of MEG/EEG data requires advanced mathematical frameworks.

Purpose of the Study:

• To extend the optical flow framework for analyzing spatiotemporal dynamics of MEG/EEG data.
• To demonstrate the mathematical well-posedness and applicability of this extended method.

Main Methods:

• Adaptation of the optical flow framework from computer vision to non-flat surfaces (scalp, cortical mantle).
• Application of regularizing constraints for estimating velocity fields.
• Quantitative evaluation of the optical flow method.

Main Results:

• Successful extension of optical flow for analyzing MEG/EEG spatiotemporal dynamics.
• Proof of concept for applying motion analysis techniques to neuroimaging data.
• Demonstration of method's well-posedness and quantitative evaluation.

Conclusions:

• The extended optical flow method offers a novel approach for exploring brain activity dynamics.
• This technique enhances the analysis of high-temporal-resolution neuroimaging data.
• The method is validated through simulations and a ball-catching paradigm analysis.