Event Abstract

Analysis of stable neural activity patterns generation and classification in neural cultures for real time robotic control

  • 1 Universidad Politécnica de Cartagena, Dpto. Electrónica, Tecnologı́a de Computadoras y Proyectos, Spain
  • 2 Universidad Nacional de Educación a Distancia (UNED), Dpto. de Inteligencia Artificial, Spain
  • 3 Instituto de Biología Molecular y Celular, Universidad Miguel Hernández de Elche, Spain

Material and methods We recorded the spiking activity from embryonic cortical cultures seeded over a multielectrode array(MEA) during spontaneous activity and after electrical stimulation for analyzing its global dynamical response. We show that the analysis of the firing rate patterns over time, estimated from a set of intrinsic features as inverse interspike intervals(IISI), average, variance, entropy, maximun value, maximun value time instant, maximun and minimun positives slopes and maximun negative slope time instant, in each electrode, and the use of machine learning techniques like ANNs can lead to distinguish stable patterns evoked by the stimulus in different electrode positions. A selected set of features together with an ensemble of artificial neural networks modelization are the procedures used in classification of the global dynamic responses. The properties of the spontaneous activity of the cultures before the stimulation experiments is measured. We will apply the stimulus over a subset of electrodes to find out which of them trigger consistent neural responses over the neural circuit. Once these electrodes have been stimulated, their evoked response is analyzed. After filtering raw data of spikes, the activity in electrodes after stimulation is compared with the basal activity (without stimulations) to detect which electrodes are sensitive to stimulations, and to discard the inactive ones. The sequence of time-stamps of the spikes obtained in the preprocessing, from the ones recorded in an active electrode during the 600 ms immediately after a stimulation, is extracted from the recording. Then, from this sequence of time-stamps a subset of features are computed. An ensemble of artificial neural networks models was built to allow robust pattern matching over the topological evoked patterns, see figure 1. The ensemble was compound of a set of artificial neural networks trained for each electrode extracted features. The networks designed has been used for classification with 4 possible categories. Decision policy of the ensemble is based in ‘winner take all’ strategy. The objective of this extensive analysis is to find a subset of electrodes that have the following properties: -The stimulation on electrode E1, E2, …, En, called ET, causes neuronal response in connected neural units recorded by receptor electrodes Ek, Ek+1, …, Ek+m, called {ER}. -The global dynamic activity (recorded from {ER}) activated by a particular ET, is different to the global dynamic activity({ER'}) activated by any other trigger electrode ET', in terms of variation over time of the firing rate estimated from a set of features in each electrode. If these conditions are fulfilled, we can assume that we have multiple independent neural patterns in the cell culture that can be evoked and classified through an ensemble of ANN models. Results The dataset has 800 patterns after preprocessing and filtering, a subset of 4 electrodes was stimulated 200 times each. The training was repeated ten times using different random weight initialization and pattern presentation order providing a satisfactory multi-pattern classification. Our best result was performed on the fourth day of experimentation with an average correct classification of (93.34±"." 98)%. For each stimulated electrodes a subset of decisive electrodes emerges. The relation between the number of times an electrode wins the decision making tournament and its reliability are shown in figure 2. Although the experiments were done off-line, all the computations are simple and could be optimized and adapted to run on-board a robot. The training of the ensemble of ANNs can be done previously off-board, and then the resulting weights could be frozen and translated into the robot to identify neuronal culture response on-line. Discussion and Conclusion There exists evidence about neurons firing properties that indicate that pulse trains are both, analogical (duration of the train pulse) and digital (“all or nothing” behavior) von Neumann (1952), as well as they manage information changes depending on the density of the train pulses. These properties suggest a strong dependency of the neural code on the frequency domain and its variation over time. The approximation presented here can take advantage of these properties to obtain a better characterization of the neuronal culture response, thus making it more useful for culture embedding in robots. Also, this method allows to explore and extend the possible forms to use the global response in a neuronal culture, because it does not depend on the response at specific electrodes after stimulation, but on parallel computational processes of the neurons connected in the neuronal culture. The results show that the neural code can be modified by local modulating activities when the correct electrode is stimulated, and the resulting neural flow feature properties can be used for this characterization.

Figure 1
Figure 2

Acknowledgements

We want to acknowledge to Programa de Ayudas a Grupos de Excelencia de la Región de Murcia, from Fundación Seneca, Agencia de Ciencia y Tecnología de la Región de Murcia. We are gratefull to Profs. Miguel Ángel Fernández Graciani and Jose María López Valles(Univ. Castilla-La Mancha, Spain).

References

-DeMarse, T. B., Wagenaar, D. A., Blau, A. W., and Potter, S. M. (2001). The neurally controlled animat: biological brains acting with simulated bodies. Autonomous Robots 11, 305–310. doi:10.1023/A: 1012407611130. -Novellino, A., D’Angelo, P., Cozzi, L., Chiappalone, M., Sanguineti, V., and Martinoia, S. (2007). Connecting neurons to a mobile robot: An in vitro bidirectional neural interface. Computational Intelligence and Neuroscience 2007. doi:10.1155/2007/12725. -Tessadori, J., Bisio, M., Martinoia, S., and Chiappalone, M. (2012). Modular neuronal assemblies embodied in a closed-loop environment: toward future integration of brains and machines. Frontiers in neural circuits 6. doi:10.3389/fncir.2012.00099. -Tessadori, J., Venuta, D., Kumar, S. S., Bisio, M., Pasquale, V., and Chiappalone, M. (2013). Embodied neuronal assemblies: A closed-loop environment for coding and decoding studies. In 6th International IEEE/EMBS Conference on Neural Engineering (NER), 2013. 899–902. doi:10.1109/NER.2013.6696080. - Warwick, K., Xydas, D., Nasuto, S. J., Becerra, V. M., Hammond, M. W., Downes, J. H., et al. (2010). Controlling a mobile robot with a biological brain. Defence Science Journal 60, 5–14. doi:10.14429/dsj.60.11 -Valentino Braitenberg. Vehicles Experiments in Synthetic Psychology. The MIT Press, 1986. -von Neumann, J. (1952). Probabilistic logics and the synthesis of reliable organisms from unreliable components. Lectures delivered at the California Institute of Technology. Notes by R. S. Pierce

Keywords: MEA, CNN, Close-loop, IISI, Robot-control

Conference: MEA Meeting 2018 | 11th International Meeting on Substrate Integrated Microelectrode Arrays, Reutlingen, Germany, 4 Jul - 6 Jul, 2018.

Presentation Type: Poster Presentation

Topic: Neural Networks

Citation: Calvo MV, Alegre-Cortés J, Lozano A, De La Paz López F, Álvarez-Sánchez J, Ferrandez JM and Fernandez E (2019). Analysis of stable neural activity patterns generation and classification in neural cultures for real time robotic control. Conference Abstract: MEA Meeting 2018 | 11th International Meeting on Substrate Integrated Microelectrode Arrays. doi: 10.3389/conf.fncel.2018.38.00106

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: 18 Mar 2018; Published Online: 17 Jan 2019.

* Correspondence: PhD. Mikel V Calvo, Universidad Politécnica de Cartagena, Dpto. Electrónica, Tecnologı́a de Computadoras y Proyectos, Cartagena, Spain, Spain, mikel1982mail@gmail.com