PROJECT TITLE :

Sparse Representation For Brain Signal Processing A Tutorial On Methods And Applications - 2014

ABSTRACT:

In many cases, observed brain signals will be assumed because the linear mixtures of unknown brain sources/elements. It is the task of blind supply separation (BSS) to seek out the sources. But, the amount of brain sources is usually larger than the amount of mixtures, which results in an underdetermined model with infinite solutions. Beneath the cheap assumption that brain sources are sparse inside a website, e.g., in the spatial, time, or time-frequency domain, we tend to could get the sources through sparse illustration. As explained in this article, several other typical issues, e.g., feature choice in brain signal processing, will additionally be formulated because the underdetermined linear model and solved by sparse illustration. This article 1st reviews the probabilistic results of the equivalence between 2 necessary sparse solutions?the 0-norm and one-norm solutions. In sparse illustration-based brain part analysis including blind separation of brain sources and electroencephalogram (EEG) inverse imaging, the equivalence is connected to the recoverability of the sources. This article additionally focuses on the applications of sparse illustration in brain signal processing, including elements extraction, BSS and EEG inverse imaging, feature choice, and classification. Based mostly on purposeful magnetic resonance imaging (fMRI) and EEG knowledge, the corresponding strategies and experimental results are reviewed.