Multi-level multi-scale tomographic method based on Bayesian estimation

Abstract

The invention discloses a multi-level multi-scale tomographic method based on Bayesian estimation. The method comprises for complicated brain source activities of different scales, decomposing abrainsource space into mutually independent voxels, wherein each voxel is used as a potential source of activity; then decomposing all the voxels into regions according to the anatomical structure or functional area of the brain, wherein respective regions correspond to different distribution parameters; in addition, analyzing the out-of-brain sampling data by using a method based on Bayesian and its convex function theory, and estimating the intrinsic activity distribution of all voxels and the covariance components of regional activity distribution. Based on the above model framework, two methodsare provided, wherein one method is that the covariance component of the final activity of the voxels is determined only by the regional activity and is called tling-Champagne, and the other method is that the activity of the voxels is determined by the intrinsic activity of the voxels and the regional activity distribution and is called tree-Champagne. Finally, the performance analysis is carried out by simulation data and real brain data, and good effects are obtained.

Description

technical field [0001] The invention belongs to the interdisciplinary field of brain science and information technology, and relates to a method for imaging magnetoencephalography (MEG) and electroencephalogram (EEG), in particular to a multi-level and multi-scale layer based on Bayesian estimation Analytical imaging method. Background technique [0002] The imaging of human brain activity plays an important role in the research of cognitive neuroscience, which can promote the understanding of the working mechanism of neurons in the complex activities of human beings. In clinical medical applications, brain imaging also plays an important role, especially Diagnosis, guidance and resection of brain diseases such as brain tumors and epilepsy surgery. However, current brain source localization and activity time-series estimation is an extremely challenging problem because the number of potential brain activity sources can be in the tens of thousands, while the number of extra-...

AI Technical Summary

Problems solved by technology

The present invention mainly aims at the problem that the current tomographic method cannot accurately image multi-scale source activities, and combines the brain functional structure and anatomical structure to segment the source space, and proposes a multi-scale source activity tomography method based on Bayesian

In recent years, many scholars have analyzed brain source activities through MEG and EEG, and proposed many methods of brain electromagnetic source imaging, but for activities such as a cluster of multiple adjacent dipoles or a cluster of multiple adjacent dipoles The problem of the simultaneous activity of sub and independent dipoles has not been able to find an effective solution

Images