Bayesian EEG source localization using a structured sparsity prior

详细信息    查看全文

• 作者：Facundo Costa ; ^a ; ^{facundo.costa@enseeiht.fr} ; Hadj Batatia^a ; Thomas Oberlin^a ; Carlos D'Giano^b ; Jean-Yves Tourneret^a
• 关键词：EEG ; MCMC ; Inverse problem ; Source localization ; Structured-sparsity ; Hierarchical Bayesian model ; ℓ_{20 norm regularization} ; Medical imaging
• 刊名：NeuroImage
• 出版年：2017
• 出版时间：1 January 2017
• 年：2017
• 卷：144
• 期：part_PA
• 页码：142-152
• 全文大小：3157 K
• 卷排序：144

文摘

A Bayesian model for solving the EEG source localization problem is proposed. The model promotes sparsity using a multivariate Bernoulli Laplacian prior. A Gibbs sampler is used to generate samples according to the posterior distribution. Metropolis–Hastings moves are used to improve the convergence rate of the sampler. The method is compared with synthetic and real data to state of the art algorithms.