Multiple Constrained Minimum-Variance Beamformers for Reconstruction of Induced and Evoked Neural Activity

Abstract

AbstractLinearly constrained minimum variance beamformers are highly effective for analysis of weakly correlated brain activity. When correlations become significant, their performance drastically degrades. One solution to this problem is to use a multiple constrained minimum variance (MCMV) beamformer which however requires a priori information about the correlated sources locations. Besides the question whether unbiased estimates of source positions can be obtained remained unanswered. In this work, we derive MCMV-based source localizers of two kinds: a) those that can be applied to both induced and evoked activity; and b) those targeting only the evoked component. We show that for arbitrary noise covariance both yield simultaneous unbiased estimates of multiple source positions and orientations and remain bounded at singular points. We also derive a search algorithm that may be used to find multiple unknown correlated sources. Simulations and analyses of real MEG data demonstrate that in situations where a single-source beamformer is unable to detect coherent sources, the MCMV-based ones are able to find either most or all of them, depending on the SNR.Keywordsmagnetoencephalography (MEG)inverse solutionsbeamformerscorrelated sources