Event Abstract

Connecting insect brain imaging to data mining tools - a neuroimage extension for the KNIME platform.

  • 1 University of Konstanz, Germany
  • 2 University of Konstanz, Department of Neurobiology, Germany

The insect antennal lobe (AL) is a brain compartment dedicated to the processing of olfactory stimuli. Spatio-temporal activity patterns in the AL, that can be recorded with calcium-imaging techniques, encode the identity of odorants.
We employ machine learning and data mining approaches in our efforts to understand the "olfactory code" and to infer the structure of the AL network from recordings. Therefore we have designed a collection of neuroimage software tools bundled as an extension package for the KNIME data exploration platform.
KNIME (www.knime.org) follows a modular concept of nodes that each implement a specific algorithm or function and can be connected to create user-defined data pipelines. KNIME offers convenient ways of data handling, features a connection to R, supports the WEKA library of machine learning algorithms etc.
Our software tools connect insect brain imaging to the KNIME environment, taking raw movies recorded in calcium-imaging experiments as input and processing them automatically. The processing comprises the complete pipeline from stabilising shaky recordings and normalisation up to pattern extraction and visualisation.
PCA, ICA and variants thereof have proven to be useful for dimensionality reduction and denoising in this environment [1]. We employ these methods for pattern extraction from AL recordings [2], allowing us to detect even faint signals in background activity [3].
The output can be visualised as time-series and false-color coded movies of activity changes, 2D maps of the AL and 3D projections onto a model of the AL. Furthermore, the extracted patterns are now directly accessible for classification and data mining approaches offered by the KNIME platform and its extensions.

References

1. Reidl, Starke, Omer, Grinvald & Spors, Neuroimage, 34(1):94-108, 2007

2. Strauch & Galizia, Lecture Notes in Informatics, P-136, pp. 85-95, 2008

3. Rein, Strauch & Galizia, poster abstract, p. 958, Proceedings of the 8th Göttingen Meeting of the German Neuroscience Society, 2009

Conference: Neuroinformatics 2009, Pilsen, Czechia, 6 Sep - 8 Sep, 2009.

Presentation Type: Poster Presentation

Topic: Neuroimaging

Citation: Strauch M, Lutz C, Samuelson D and Galizia GC (2009). Connecting insect brain imaging to data mining tools - a neuroimage extension for the KNIME platform.. Front. Neur. Conference Abstract: Neuroinformatics 2009. doi: 10.3389/conf.neuro.11.2009.08.020

Received: 21 May 2009; Published Online: 21 May 2009.

* Correspondence: Martin Strauch, University of Konstanz, Konstanz, Germany, Martin.Strauch@lfb.rwth-aachen.de

© 2007 - 2018 Frontiers Media S.A. All Rights Reserved