Premium
Functional source separation and hand cortical representation for a brain–computer interface feature extraction
Author(s) -
Tecchio Franca,
Porcaro Camillo,
Barbati Giulia,
Zappasodi Filippo
Publication year - 2007
Publication title -
the journal of physiology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.802
H-Index - 240
eISSN - 1469-7793
pISSN - 0022-3751
DOI - 10.1113/jphysiol.2007.129163
Subject(s) - brain–computer interface , magnetoencephalography , computer science , electroencephalography , blind signal separation , brain activity and meditation , source separation , set (abstract data type) , interface (matter) , artificial intelligence , exploit , representation (politics) , signal (programming language) , pattern recognition (psychology) , channel (broadcasting) , neuroscience , psychology , bubble , maximum bubble pressure method , parallel computing , politics , political science , law , computer network , computer security , programming language
A brain–computer interface (BCI) can be defined as any system that can track the person's intent which is embedded in his/her brain activity and, from it alone, translate the intention into commands of a computer. Among the brain signal monitoring systems best suited for this challenging task, electroencephalography (EEG) and magnetoencephalography (MEG) are the most realistic, since both are non‐invasive, EEG is portable and MEG could provide more specific information that could be later exploited also through EEG signals. The first two BCI steps require set up of the appropriate experimental protocol while recording the brain signal and then to extract interesting features from the recorded cerebral activity. To provide information useful in these BCI stages, our aim is to provide an overview of a new procedure we recently developed, named functional source separation (FSS). As it comes from the blind source separation algorithms, it exploits the most valuable information provided by the electrophysiological techniques, i.e. the waveform signal properties, remaining blind to the biophysical nature of the signal sources. FSS returns the single trial source activity, estimates the time course of a neuronal pool along different experimental states on the basis of a specific functional requirement in a specific time period, and uses the simulated annealing as the optimization procedure allowing the exploit of functional constraints non‐differentiable. Moreover, a minor section is included, devoted to information acquired by MEG in stroke patients, to guide BCI applications aiming at sustaining motor behaviour in these patients. Relevant BCI features – spatial and time‐frequency properties – are in fact altered by a stroke in the regions devoted to hand control. Moreover, a method to investigate the relationship between sensory and motor hand cortical network activities is described, providing information useful to develop BCI feedback control systems. This review provides a description of the FSS technique, a promising tool for the BCI community for online electrophysiological feature extraction, and offers interesting information to develop BCI applications to sustain hand control in stroke patients.