Biomedical Engineering Reference
In-Depth Information
Chapter 3
Mining Spatio-Spectro-Temporal Cortical
Dynamics: A Guideline for Offl ine and Online
Electrocorticographic Analyses
Mining Cortical Dynamics from ECoG Data
Zenas C. Chao and Naotaka Fujii
Abstract
Recent advances in the technology of electrocorticography (ECoG)
allow accessing neural activity from most of the cortex, which poses the challenge
of extracting relevant information from an overwhelming amount of data. In this
chapter, we will present useful routines for identifying statistically signifi cant fea-
tures in high-dimensional ECoG signals (offl ine analysis) and for establishing
decoding models that can translate ECoG signals to specifi c behavioral measures in
real time (online analysis). We will use our data, which are freely available online,
in a step-by-step demonstration and will highlight useful MATLAB toolboxes for
trouble-free implementation.
Keywords
Connectivity • Cortical dynamics • Data mining • Electrocorticography
(ECoG) • MATLAB • Offl ine analysis • Online analysis • Time-frequency
representation
3.1
Introduction
Electrocorticography (ECoG) records brain activity using grids or stripes of
electrodes implanted on the surface of the cortex. As most large cortical neurons are
oriented perpendicular to the cortical surface, correlated activity within a cortical
column should sum similarly. Consequently, ECoG is most favorable, among the
brain activity recording techniques available, for measuring this correlated activity,
especially across a wide area.
