The Study Of Electroencephalogram Inverse Problem By Data Of Spatio-Temporal

The study of electroencephalogram (EEG) includes forward solution of EEG and inverse problem of EEG. An ocular method about solving inverse problem is to optimizing parameters of forward problem. That is constructing objectives. By adopting definite non-linear optimization algorithm, it can iterative the algorithm and then to approach the objectives. These are all models of equivalent dipole based on a single slice of time. As introducing the model of spatio-temporal, it help to reduce the complexity of non-linear optimization. And some methods in modern signal disposal had been applied to the study of EEG and MEG inverse problem. By using the MUSIC algorithm to locate multiple dipolar sources from EEG and MEG data, it can avoid the local minimum of non-linear least square. The algorithm can locate multiple dipolar source by scanning a single-dipole model through the whole volume. Subspace projection as a method of signal disposal has been recently used in EEG and MEG inverse problem. With it, we can eliminate the effects of non objected sources in the signals or by eliminate false trace to increase the precision of source location. Besides these, this article propose to use recursive algorithm and use the extracted matrix and the method of projection of subspace. The signal subspace method is to use the found source to form a matrix and then projected subspace into its orthogonal complement. This method can reduce complexity. Using extracted matrix and gain matrix to compute the value of subspace correlations can improve the veracity. And last by using recursive algorithm can found asynchronous and synchronous sources. And experiments prove that this method can find the synchronized dipoles more exactly than the pure MUSIC algorithm.