Multivariate analysis of the activity of the olfactory bulb

In the understanding of processes of neural activity in complex networks, non-invasive recording of the electrical activity is desirable. One method that achieves this is through the use of voltage-sensitive fluorescent dyes (VSD) as reporters, which convert changes in a tissue's membrane potential into fluorescent emission. This fluorescent activity is recorded by means of fast CCD cameras that allow visualization and a posteriori studies. Image sequences obtained in this way, although often noisy, are commonly studied following a univariate approach. In this work, there are studied several series of image sequences from an experiment initially focused in the monitoring of the Olfactory Bulb activity of a frog (Rana temporaria), where the olfactory receptors were exposed to two volatile compounds, under different inhibitory conditions. Our work proposes the use of two multivariate analysis methods such Multi-way Principal Component Analysis and Independent Component Analysis with a dual aim. First, they are able to improve the recordings by removing noise and aliasing after using a supervised selection of parameters. Secondly, they demonstrate possibilities in the obtaintion of simultaneous information about the most active areas of the monitored surface and its temporal behaviour during the stimulus. © 2009 Springer-Verlag Berlin Heidelberg.