Nonsurgical Brain Activity Recovery From a Cap Containing Multiple Electroencephalogram Recording Sites

摘要

The goal of this project is to invert electroencephalogram data to brain activity using a novel and accurate algorithm. This algorithm can be used for research and development of brain-machine interfaces, potentially to include prosthetic control, surveillance and lie detection. This report details the mathematical and physical theory behind the algorithm developed during this effort. The algorithm is novel in that it does not use the quasistatic approximation throughout the space, instead using it only at the boundaries between material media. The output of the algorithm consists of three- dimensional currents at dipoles located at pre-specified locations. The number of such dipoles is limited to one third of the number of electroencephalogram electrodes, so they are best regarded as neuronal aggregates instead of individual neurons. The algorithm has been executed on up to five seconds of 128-electrode electroencephalogram data sampled at 2048 Hertz. Numerical results are provided in graphical format.