Supporting fMRI Data Analysis by Structured Neural Networks
-
1
Leibniz Institute for Neurobiology, Germany
-
2
Otto-von-Guericke University, Germany
The aim of our work in the last years was to find an artificial neural network structure for the analysis of time series which can be used in a better way than conventional statistical methods. The analysis of stimulus related data is a common investigation method in neuroscience. Any improvement of analytical methods can be helpful for better understanding the complexity of the brain. In this sense also alternatives to common statistical methods, like the proposed structured neural network, should be investigated, especially their usefulness in respect to the stimulus related data analysis, exemplified by fMRI studies with auditory stimuli.
The new attempt to the data analysis published in [1] is a combination of a spiking network, the Liquid State Machine (LSM), with a self-organizing map (Multilevel Hypermap Architecture, MHA) as a readout. The neuron model of Izhikevich [2] is adapted for using as a LSM. LSM’s ([3]) were conceived from a mathematical and computational neuroscience perspective and usually are made of biologically more plausible, spiking neurons with a continuous-time dynamics. The Multilevel Hypermap Architecture (MHA) is classified under self-organizing neural networks and is an extension of the Hypermap introduced by Kohonen. The MHA supports multi-level data vectors, therefore the MHA is useful for the analysis of structured or hierarchical data. An overview of our last works about MHA gives [4].
The classification of the fMRI data with the LSM-MHA shows similar results in comparison to the statistical tests, but in average with an improvement of discrimination of about 15 percent. The results of our analysis of fMRI data sets by means of the LSM-MHA show, that it is possible to analyze such periodically structured data. Furthermore is the LSM-MHA an useful complement to conventional statistical methods in this field. Because of its character like a simultaneous auto-correlation and cross-correlation to the stimulus it has a higher selectivity and discrimination than statistical methods. Therefore the LSM-MHA should be also a preferred tool for the analysis of other kinds of stimulus related data, like event related potentials (ERP).
With the implementation of the MHA algorithm in MATLAB and the integration of an interface to BrainVoyager ([5]) it is possible to visualize the classified data in VOI’s.
References
1. Brückner, B.: Stimulus related data analysis by structured neural networks. In: Lecture Notes in Computer Science. Number 4234/2006, Springer-Verlag (2006) 251–259
2. Izhikevich, E.M.: Simple model of spiking neurons. IEEE TRANSACTIONS ON NEURAL NETWORKS 14(6) (2003) 1569–1572
3. Maass, W., Natschläger, T., Markram, H.: Real-time computing without stable states: A new framework for neural computation based on perturbations. Neural Computation 14(11) (2002) 2531–2560
4. Brückner, B., Wesarg, T.: Modeling speech processing and recognition in the auditory system using the multilevel hypermap architecture. In Seiffert, U., Jain, L., eds.: Self-Organizing Neural Networks. Recent Advances and Applications. Volume 78 of Springer Series on Studies in Fuzziness and Soft Computing., Heidelberg, Springer-Verlag (2001) 145–164
5. (www.brainvoyager.com)
Conference:
Neuroinformatics 2009, Pilsen, Czechia, 6 Sep - 8 Sep, 2009.
Presentation Type:
Poster Presentation
Topic:
Neuroimaging
Citation:
Brückner
B and
Walter
T
(2019). Supporting fMRI Data Analysis by Structured Neural Networks.
Front. Neuroinform.
Conference Abstract:
Neuroinformatics 2009.
doi: 10.3389/conf.neuro.11.2009.08.029
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:
21 May 2009;
Published Online:
09 May 2019.
*
Correspondence:
Bernd Brückner, Leibniz Institute for Neurobiology, Magdeburg, Germany, brueckner@ifn-magdeburg.de