Information Quantification for Spike Trains and Field Potentials
Zareen Mehboob · Research Explorer (The University of Manchester) · 2011
The University of ManchesterZareen Mehboob2010Degree of Doctor of PhilosophyNeural signals are recorded from various regions of the brain and are analysed tounderstand the working mechanism of neurons and how they interpret externalenvironment. The aim is to understand how this nature?s supercomputer works.This helps in exploring human systems and intelligence, treat mental conditionsand develop smart machines. Neural data recordings are collected from individualneurons and from populations of neurons. The single neuronal activity recordingsare spike train and the activity generated from multiple neurons are field poten-tials. The data obtained are in enormous amount and of millisecond precision,as a consequence their processing is not a trivial task and efficient techniques arerequired for decoding these datasets.This work proposes several methods for the analysis of spike train and fieldpotentials. A self-organising map based clustering is applied to synchronous spiketrain and generates topographically ordered and information-preserving clustersthat help interpret how stimuli features are encoded by the neurons.An information-coupled empirical mode decomposition framework is devel-oped for field potentials. It extracts informative oscillatory functions and infor-mation coding frequency bands in the recordings. This has several applications.The informative modes reveal underlying neuronal activities w.r.t stimuli, whichotherwise have to be extracted by bandpass filters, followed by Fourier or waveletsanalysis. It can also be used to analyse neuronal population activity under a med-ical condition or to understand neuronal interactions by information-connectivityanalysis among electrodes. The proposed framework is developed into the formof a toolbox which can be used for educational and research purposes.