Connecting biological and artificial neural networks
-
1
Istituto Italiano di Tecnologia, Department of Neuroscience and Brain Technologies, Italy
-
2
University of Bordeaux, Laboratoire IMS, France
-
3
Istituto Italiano di Tecnologia, Department of Neuroscience and Brain Technologies, Italy
Motivation
Neurological disorders can disrupt connections among different brain regions, resulting in severe damage to cognitive and motor capabilities [1]. In the near future, neural stimulators and implantable systems represent one of the most promising technologies to reduce those impairments [2]. This work is part of the European project BRAIN BOW (www.brainbowproject.eu) whose purpose was to build a test-bed for the development and the study of a new generation of neuro-prostheses capable to restore the communication between neuronal circuitries lost because of a neural lesion, such as a traumatic brain injury [3].
Material and Methods
The biological element used in this study is constituted by dissociated cortical rat neurons plated over Micro-Electrode Arrays (MEAs). We adopted a neuromorphic board which is able to perform real-time event detection and trigger an electrical stimulation of the Biological Neural Network (BNN). The board (figure 1, panel A) embeds an Artificial Neural Network (ANN) [4], based on Izhikevich neurons [5] which can be put in uni- and bi-directional communication with the BNN. The ANN used in the following experiments was made up of 100 neurons (80 excitatory, 20 inhibitory) randomly connected with an average degree of 45.
Results
We tested the effect of stimulation from the BNN delivered to the first 1-2-5-10 excitatory neurons in the ANN. The ANN receives stimulation each time a spike is detected on a channel of the BNN. In figure 1, panel B, we show the probability of observing a spike on one of the target channels of the ANN as a function of time occurring since a stimulus was delivered to it. It is possible to note how this graph presents two well-distinct bumps. The highest is almost overlapping in all curves and spans between 5 and 15 ms after stimulus delivery. This peak is likely caused by the stimulated neurons responding with a spike to the external stimulation directly. A delayed bump is present only when 10 excitatory neurons are stimulated, clearly visible in the 20-40 ms time interval after stimulation. This peak is caused by the fact that if a large enough number of excitatory neurons is firing at the same time, the network generates a self-sustaining barrage of activity. Both these peaks are typically observed in biological cultures undergoing electrical stimulation, with similar time profiles (even though evoked bursts tend to last longer, up to hundreds of ms [6]).
As expected, when inhibitory rather than excitatory neurons are used as stimulation targets, the only macroscopic difference is the lack of the rightmost peak, i.e. the late network response evoked by stimulation (data not shown).
Discussion
Here we show that the activity of the ANN, implemented into a neuromorphic board, can be influenced by the BNN. With this particular configuration of ANN, we have to stimulate at least 10 artificial excitatory neurons in order to cause a network response. The unidirectional stimulation from BNN to ANN can be seen as a preliminary test for the bidirectional stimulation. During bidirectional stimulation, we expect that the time profile of stimulation and the number of stimuli delivered between ANN and BNN can have a profound effect on the information exchange between networks.
Conclusion
The results (although preliminary) of this work demonstrate that the board is able to put in communication a biological and an artificial neural network.
References
[1] D. J. Guggenmos, M. Azin, S. Barbay, J. D. Mahnken, C. Dunham, P. Mohseni, et al., "Restoration of function after brain damage using a neural prosthesis," Proc Natl Acad Sci U S A, vol. 110, pp. 21177-82, Dec 24 2013.
[2] E. Dayan, N. Censor, E. R. Buch, M. Sandrini, and L. G. Cohen, "Noninvasive brain stimulation: from physiology to network dynamics and back," Nat Neurosci, vol. 16, pp. 838-44, Jul 2013.
[3] P. Bonifazi, F. Difato, P. Massobrio, G. L. Breschi, V. Pasquale, T. Levi, et al., "In vitro large-scale experimental and theoretical studies for the realization of bi-directional brain-prostheses," Front Neural Circuits, vol. 7, p. 40, 2013.
[4] M. Ambroise, T. Levi, S. Joucla, B. Yvert, and S. Saighi, "Real-time biomimetic Central Pattern Generators in an FPGA for hybrid experiments," Front Neurosci, vol. 7, p. 215, 2013.
[5] E. M. Izhikevich, "Simple model of spiking neurons," IEEE Transactions on Neural Networks, vol. 6, pp. 1569-1572, 2003.
[6] D. A. Wagenaar, J. Pine, and S. M. Potter, "An extremely rich repertoire of bursting patterns during the development of cortical cultures," BMC Neurosci, vol. 7, p. 11, 2006.
Figure Legend
Figure 1: Panel A, picture of the neuromorphic board used for this study. Panel B, Peristimulus spiking probability on one of the target channels of the ANN as a function of time occurring since a stimulus was delivered to it.
Acknowledgements
The research leading to these results has received funding from the European Union's Seventh Framework Programme (ICTFET FP7/2007-2013, FET Young Explorers scheme) under grant agreement n. 284772 BRAIN BOW (www.brainbowproject.eu).
Keywords:
Electrical Stimulation,
MEA,
in vitro,
hybrid networks,
real-time event detection,
neuromorphic board
Conference:
MEA Meeting 2016 |
10th International Meeting on Substrate-Integrated Electrode Arrays, Reutlingen, Germany, 28 Jun - 1 Jul, 2016.
Presentation Type:
Poster Presentation
Topic:
MEA Meeting 2016
Citation:
Buccelli
S,
Tessadori
J,
Bornat
Y,
Pasquale
V,
Ambroise
M,
Levi
T,
Massobrio
P and
Chiappalone
M
(2016). Connecting biological and artificial neural networks.
Front. Neurosci.
Conference Abstract:
MEA Meeting 2016 |
10th International Meeting on Substrate-Integrated Electrode Arrays.
doi: 10.3389/conf.fnins.2016.93.00095
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:
22 Jun 2016;
Published Online:
24 Jun 2016.
*
Correspondence:
Dr. Stefano Buccelli, Istituto Italiano di Tecnologia, Department of Neuroscience and Brain Technologies, Genova, Italy, stefano.buccelli@gmail.com