Mathematical processing of in-vivo electrophysiological data with applications to implanted electrode recordings

Abstract

Neurons are the principal cellular elements that underlie the function of the nervous system, which includes the brain, spinal cord, and peripheral ganglia. These electrically excitable cells process and transmit information primarily via electrical signalling through the generation of action potentials. These action potentials can be recorded in-vivo by placing electrodes in the vicinity of the neuron’s membrane within the extracellular space. Electrodes measure electric potential fluctuations in the extracellular space. These fluctuations generally contain two types of activity, low frequency content, also known as Local Field Potentials (LFPs) and extracellular action potentials (spikes) which contribute to the higher frequency content. The advancement of neural recording techniques allows for the simultaneous recording of many neurons. It has been estimated that the ability to record simultaneously from several neurons has been growing exponentially since the 1950s and it has been predicted that this number doubles every 7 years (Stevenson and Kording, 2011). The technological advance in neural recording systems demands the parallel advancement of neural decoding algorithms to analyse and extract information from these signals automatically and objectively.