Spike sorting and functional connectivity analysis using self-organizing maps and granger causality
-
1
University of Sussex, United Kingdom
We have constructed an integrated workflow system for analyzing dynamic connectivity in data from multi-unit electrode neural recordings. Our software system is constructed using a combination of blind source separation (BSS) implemented with a Kohonen self-organizing map (SOM) and a Granger causality (GC) system for illuminating the functional connectivity of networks recorded. We present the analysis of extracellular planar multi-electrode array data from both simulated recordings and raw data collected from the feeding system of a semi-intact snail brain. SOM networks have been used for a variety of automatic tuning problems for examining source separation of time series data.
In this case we use a SOM to identify several individual neurons recorded from the electrodes in our arrays. Our recording array consists of sixty circular electrodes spaced at distances ranging from thirty microns to two hundred microns, center to center. Depending on the spacing of the array used, there is a high likelihood of the membrane potential of more than one neuron being recorded on each electrode. This creates a need to perform some form of source separation on data recorded by this system to identify and differentiate data from individual neurons in each electrode voltage trace. Spike sorting of this nature presents a formidable challenge for performing large-scale dynamic analysis of neural data collected from many neurons in a network simultaneously.
Waveform data being presented to the SOM network is preprocessed by automated thresholding to select neural spikes and then further segmentation is performed using either principal components or independent components analysis. This provides us with a number of components representing the shapes found in a particular waveform. In addition to the waveform data, we add additional weights in the training vectors to represent the physical relationship of electrodes in the array. This step provides us with waveform separation according to both location of the electrode it was recorded from and its shape components.
Separated source data is then analyzed using a Granger causality system to identify likely functional connectivity between identified neurons. This step also validates the SOM-BSS method by eliminating false multiple classification of spikes and highlights the presence of neural connections of chemical synapses and neural syncytia (groups of cells tightly coupled by electrical synapses). If we look at data using our Granger causality tools on varying time scales, we can see changes in functional connectivity of the feeding network in snails undergoing stimulation or training. We believe these varying functional maps represent the presence of different behavioral states within the system being examined, and in the case of training, may represent encoding of learning events. By using the combination of these techniques (SOM-BSS and GC) we are attempting to build dynamic activity maps of large networks of neurons recorded with multi-unit techniques.
Conference:
Neuroinformatics 2008, Stockholm, Sweden, 7 Sep - 9 Sep, 2008.
Presentation Type:
Poster and Short Oral Presentation
Topic:
Electrophysiology
Citation:
Passaro
P,
Harris
C,
Seth
A,
O'Shea
M and
Husbands
P
(2008). Spike sorting and functional connectivity analysis using self-organizing maps and granger causality.
Front. Neuroinform.
Conference Abstract:
Neuroinformatics 2008.
doi: 10.3389/conf.neuro.11.2008.01.099
Copyright:
The abstracts in this collection have not been subject to any Frontiers peer review or checks, and are not endorsed by Frontiers.
They are made available through the Frontiers publishing platform as a service to conference organizers and presenters.
The copyright in the individual abstracts is owned by the author of each abstract or his/her employer unless otherwise stated.
Each abstract, as well as the collection of abstracts, are published under a Creative Commons CC-BY 4.0 (attribution) licence (https://creativecommons.org/licenses/by/4.0/) and may thus be reproduced, translated, adapted and be the subject of derivative works provided the authors and Frontiers are attributed.
For Frontiers’ terms and conditions please see https://www.frontiersin.org/legal/terms-and-conditions.
Received:
28 Jul 2008;
Published Online:
28 Jul 2008.
*
Correspondence:
Peter Passaro, University of Sussex, Brighton, United Kingdom, p.a.passaro@sussex.ac.uk