Multiresolution MEG (magnetoencephalography) imaging by adaptive beamformer scanning

There is a tradeoff between the spatial resolution and the speed of beamformer MEG imaging method. For clinical purposes, source localization for large volumes of MEG/EEG data such as epileptic spike data needs to be performed quickly, accurately, and automatically. In an effort to meet these requirements we have developed multiresolution beamformer MEG imaging technique to localize brain electrical activity with high speed and fine resolution. To avoid scanning the areas without activity, the method iteratively refines the grid only in the areas where activity has been detected. The method produces significant savings in processing time. Numerical experiment shows 400 times speed up in comparison with conventional beamformer method.

I. Introduction

Magnetoencephalography (MEG) is a non-invasive tool capable of investigating the dynamics of neuronal activity in the living human brain. One major advantage of MEG is its ability to measure brain activity with fine temporal resolution, on the scale of milliseconds [1]. This advantage of MEG has led many researchers to develop MEG source localization techniques. But it has been a difficult task to get good solution due to the ill-posed nature of MEG source localization [2], [3], [4]. Nevertheless, several researchers reported that under certain conditions the spatial resolution of the MEG source localization can reach a few millimeters [5], [6], [7], [8].