Transfer-Function-Based Calibration of Sparse EEG Systems for Brain Source Localization

This paper describes a transfer-function-based calibration technique for a sparse electroencephalography (EEG) source localization system. The forward solution and synthetic EEG signals are generated using the Brainstorm software. Various forms of calibration errors have been defined and induced on the nominal manifold, with different magnitudes. Two transfer-function-based calibration techniques are presented and the improvement brought about using these methods are compared against the results from benchmark scenarios. The performance of the proposed calibration techniques has been quantified in terms of the localization errors in the recovered sources before and after calibrating the system. Simulations suggest that in the presence of calibration errors, the proposed technique is capable of reducing the localization error as well as the number of falsely recovered sources.