Investigating the Collective Behavior of Neural Networks: A Review of Signal Processing Approaches

Investigating the Collective Behavior of Neural Networks: A Review of Signal Processing Approaches

A. Maffezzoli (Politecnico di Milano, Italy)
Copyright: © 2009 |Pages: 15
DOI: 10.4018/978-1-60566-076-9.ch032
OnDemand PDF Download:


In this chapter, authors review main methods, approaches, and models for the analysis of neuronal network data. In particular, the analysis concerns data from neurons cultivated on Micro Electrode Arrays (MEA), a technology that allows the analysis of a large ensemble of cells for long period recordings. The goal is to introduce the reader to the MEA technology and its significance in both theoretical and practical aspects of neurophysiology. The chapter analyzes two different approaches to the MEA data analysis: the statistical methods, mainly addressed to the network activity description, and the system theory methods, more dedicated to the network modeling. Finally, authors present two original methods, introduced by their selves. The first method involves innovative techniques in order to globally quantify the degree of synchronization and inter-dependence on the entire neural network. The second method is a new geometrical transformation, performing very fast whole-network analysis; this method is useful for singling out collective-network behaviours with a low-cost computational effort. The chapter provides an overview of methods dedicated to the quantitative analysis of neural network activity measured through MEA technology. Until now many efforts were devoted to biological aspects of this problem without taking in to account the computational and methodological signal processing questions. This is precisely what the authors try to do with their contribution, hoping that it could be a starting point in an interdisciplinary cooperative research approach.
Chapter Preview


In last years Neuroscience has been greatly enriched by engineering techniques and methods, and this scientific exchange supports specific applications of micro- or nano-technology in neurobiology and molecular biology. This new high-technological approach is called “Neuroengineering”. Engineering contribution is not simply restricted to instrumentation, but it also supplies various approaches to analyze neuronal activity, studying mathematical models and computer aided simulation of neurobiological phenomena present in “in vitro” and “in vivo” cultivations.

The widespread instrumentation is the, so-called, Micro Electrode Array (MEA) technology complementing traditional electrophysiological techniques in neuroscience research (e.g., how the brain stores and process information), prosthesis development (using living neurons as components of an integrated circuit or directly connecting a computer to them), and bio-analytics and information technology. MEA technology is very helpful to understand the dynamics of a functioning neuronal network, because it allows understanding of which different processes or components are acting together at the same time, going beyond the traditional single-neuron approach.

Engineering contribution is not however restricted to instrumentation design, and the related utilization, but it supplies advanced approaches for the neuronal activity investigation; among others, we recall data recording and elaboration, mathematical modelling and computer aided simulations of neurobiological phenomena, with virtual simulations of the behavior of a single neuron and cluster of neurons both “in vitro” and “in vivo” cultivations.

At this aim, neuroengineering gives extremely valid tools to get all information coming from neurons in an extremely wide scale, from system behavior down to single neuron. MEA data can be elaborated with custom methods of signal processing and pattern recognition or machine learning.

Specific and very promising long-term applications of MEA technologies are chemicals and pharmaceuticals set-ups, where new drugs are tested on “in vitro” neuron ensemble. At this aim, a tool able to implement a method for the evaluation of neuronal network behavior, as a consequence of different stimuli, would be is very useful by making such tests smarter and cheaper.

MEA instrumentation, summing up, complements traditional electrophysiological techniques for:

  • Fundamental neuroscience research

  • in-vitro drug assays

  • Bio-analytics (biosensors)

  • Prosthesis development

  • Information technology

    • and allows:

  • Long-term cultures

  • Multi-site extra-cellular recordings/stimulations

  • Combination with micro-fluidic systems and bio-patterning techniques

Key Terms in this Chapter

Spikes: An action potential or spike is an electro-chemical discharge traveling along the membrane of a cell, rapidly carrying information within and between neurons and indeed tissues. An action potential is a rapid change of the polarity of the voltage from negative to positive and then back to negative, the entire cycle lasting on the order of milliseconds. This cycle shows a rising phase, a falling phase, and finally an undershoot. After spiking episodes cells are unable to spike for a time called refractivity period; usually such phenomenon holds for 2-5 ms, depending from neurons types.

Slice Cultures: Slice cultures are neuron cultures in which existing connections are not removed (compare with: Dissociated cultures). A typical set-up is the analysis of mice hippocampus; it is sectioned in slices which, when placed on MEA devices, allow the analysis of mature connections.

MEA: A microelectrode array (MEA) is an arrangement of several, typically 64, electrodes allowing the targeting of several sites for stimulation and extracellular recording at once.

Long-Term Depression (LTD): LTD in neurophysiology consists in the weakening of a neuronal synapse activity obtained by prolonged low frequency stimulation on pre-synaptic neuron. It is another feature of neuronal plasticity phenomena, as LTP, and it is considered involved in learning and memory formation processes.

Long-Term Potentiation (LTP): LTP consists in an increase of the strength of chemical synapsis both in experimental preparations ( in vitro) and in living animals ( in vivo ). It is stimulated by applying a series of short, high-frequency electric stimuli on pre-synaptic neuron, able to potentiate the synapsis for minutes to hours. LTP is involved in synaptic plasticity in living animals, providing the foundation for a highly adaptable nervous system, and so in memory formation and behavioral learning. LTP was discovered in the mammalian hippocampus by Terje Lømo in 1966.

Dissociated Cultures: Dissociated cultures are cultures in which neurons, taken from an already formed brain, are chemically and mechanically treated in order to remove existing connections, and then placed on MEA devices, allowing the analysis of synaptogenic processes.

Complete Chapter List

Search this Book: