Using single/multi-channel energy transform as preprocessing tool for magnetoencephalographic data-based applications

Abstract

The purpose of this preliminary work is to evaluate the effectiveness of the single/multi-channel energy transform (ET) as preprocessing tool for magnetoencephalographic (MEG) data-based applications. The ET is a derivative-based transformation that enhances either the variability content of a signal from a single channel, or the compound variability content of signals from multiple channels. In the case of the single-channel ET, a spatial focusing effect in MEG data is achieved, which is a desirable effect given that MEG spatial variability can be correlated to regions of brain activity. On the other hand, when the ET is applied to channels that have been grouped following certain anatomical or physiological criteria, the variability content of the group gets concentrated and a signal compression is achieved. This effect can be useful in MEG-based brain-computer interfaces (BCI) where channel density compression is desired when going from training data to real life applications. Both the spatial focusing and compression properties of the ET are demonstrated with real MEG experiments.