Brain stimulation techniques with the aid of nonlinear delayed neurofeedback and MEG inverse methods

Abstract

In order to apply model-based neurofeedback methods for the regulation of brain rhythms, an appropriate biomagnetic inverse routine has been developed. This technique has been implemented in a realtime system which processes magnetoencephalography (MEG) signals for the feedback. Linear and nonlinear delayed feedback protocols have been investigated in an MEG experiment with visual stimulation. The superposition of neural activity from different brain areas in the feedback signal is unavoidable in MEG. The influence of this MEG characteristic on nonlinear delayed feedback has been studied in a model of interacting neural ensembles. A benchmark of common distributed source models for the reconstruction of the current density distribution in the brain from MEG data has been performed. For this, the reconstruction of single and multiple current dipoles has been simulated varying their position and orientation across the source space. The presence of detector noise has been taken into account. With respect to realtime applications a weighted Minimum Norm method called GaussMN, which has been tested here against other state-of-the-art algorithms for the first time, showed the best accuracy. A useful regularization routine has been developed. A spatial filter system has been produced to reduce the cross-contamination from different brain regions in the reconstructed current densities. In numerical simulations it exhibited a significantly better performance than alternative techniques. A nonlinear delayed feedback method recently proposed for the regulation of brain rhythms is investigated in interacting ensembles of coupled limit cycle oscillators. With respect to the desynchronization and decoupling of two interacting populations, an adequate mixing of their mean fields for the calculation of the stimulation signal leads to an enhanced effect. This is explained by studying the induced phase shift between the synchronized mean fields. A modeling approach for an analytical understanding is given. The realtime current density reconstruction is verified in phantom experiments. An MEG experiment applying visual delayed neurofeedback in a healthy subject has been performed. The nonlinear delayed feedback method shows a strong suppression of a brain rhythm in this proof of principle. Using more localized brain activity for the feedback signal enhances the effect.