Multi-electrode array recording and data analysis methods forudmolluscan central nervous systems

摘要

In this work the use of the central nervous system (CNS) of the aquaticudsnail Lymnaea stagnalis on planar multi-electrode arrays (MEAs) wasuddeveloped and analysis methods for the data generated were created.udA variety of different combinations of configurations of tissue from theudLymnaea CNS were explored to determine the signal characteristicsudthat could be recorded by sixty channel MEAs. In particular, theudsuitability of the semi-intact system consisting of the lips, oesophagus,udCNS, and associated nerve connectives was developed for use onudthe planar MEA. The recording target area of the dorsal surface ofudthe buccal ganglia was selected as being the most promising for studyudand recordings of its component cells during fictive feeding behaviourudstimulated by sucrose were made. The data produced by this type ofudexperimentation is very high volume and so its analysis required theuddevelopment of a custom set of software tools. The goal of this tooludset is to find the signal from individual neurons in the data streams ofudthe electrodes of a planar MEA, to estimate their position, and thenudto predict their causal connectivity. To produce such an analysis techniquesudfor noise filtration, neural spike detection, and group detectionudof bursts of spikes were created to pre-process electrode data streams.udThe Kohonen self-organising map (SOM) algorithm was adapted forudthe purpose of separating detected spikes into data streams representingudthe spike output of individual cells found in the target system. Audsignificant addition to SOM algorithm was developed by the concurrentuduse of triangulation methods based on current source densityudanalysis to predict the position of individual cells based on their spikeudoutput on more than one electrode. The likely functional connectivityudof individual neurons identified by the SOM technique were analysedudthrough the use of a statistical causality method known as Grangerudcausality/causal connectivity. This technique was used to produce audmap of the likely connectivity between neural sources.