Signal-space projection method for separating MEG or EEG into components

Abstract

The mathematical basis of signal-space projection (SSP) method was related to other comparable methods and the accuracy of the method was analyzed. Projection operators were constructed to divide the electroencephalography (EEG) and magnetoencephalography (MEG) signals into two parts. Errors were inversely proportional to the signal-to-noise ratio and sin Θ. The directions of the signals, the locations and orientations of the dipoles included in the projection operator were assumed to be known. The error ΔΓ in knowing the exact directions in the signal space caused additional noise in the analysis. The angle Θ between component vectors determined the separability of the two sources.