Closed-loop neurohybrid interfaces: from in vitro to in vivo studies - PowerPoint PPT Presentation

6 th International Workshop on Hybrid Systems and Biology - HSB’19 April 6-7, 2019 - Prague (Czech Republic) Closed-loop neurohybrid interfaces: from in vitro to in vivo studies and beyond Michela Chiappalone, PhD Rehab Technologies Istituto Italiano di Tecnologia Genova, Italy

2

Genova & IIT Central Labs 3

Disclaimer I am not a Computer Scientist I am a Biomedical Engineer… so maybe not even a ‘real’ engineer! 4

What is a hybrid system? 5

From hybrid to neuro hybrid 6

Why is it important to develop neurohybrid systems? Neuroscience  Basic Neuroscience Neuroengineering  Brain Repair  Neurorehabilitation  …  Wetware Technology  New computational paradigms Computer Science AI Robotics 7

Diseases and injuries of the central nervous system affect more than one billion people worldwide 8

‘Closed - loop’ neurohybrid interfaces connecting neuronal and artificial systems can be used to fix the brain  Brain Modulators (ICMS, DBS, NIBS)  Neuroprosthetics &  BCI/BMI Neurorobotics

Brain modulators  Replacing pharmaceutical interventions by The ‘ electroceutical ’ concept: targeted electrical stimulation delivered by smart microfabricated devices ‘A jump start for electroceuticals’, Nature April 2013 ‘Electroceuticals spark interest’, Nature July 2014 10

BCI/BMI  Neural signals are recorded from the cortex using scalp or intra-cortical electrodes. Specific features are extracted from the signals (e.g. amplitudes of evoked potentials or sensorimotor cortex rhythms, firing rates of cortical neurons). The features are then translated into a pattern of commands for an output device (e.g. a simple word processor, a robot arm, a robotized wheelchair). …01011....… DECODING ENCODING ( 𝑦, ሶ 𝑦, ሷ 𝑦 ) 11

Invasive Brain Machine Interfaces - BMIs Chapin et al. Nature Neurosci , 1999; Wessberg et al. Nature 2000; Action from thoughts , MAL Nicolelis, Nature 2001 Neural signals recorded from the brain as input commands to control external devices Hochberg et al. Nature , 2012 Bensmaia & Miller, J. Donoghue’s lab at Brown University Hochberg et al. Nature , 2006 Nat Rev Neurosci , 2014 First implants on human subjects 12

Neuroprosthetics “ a device or system that has an interface with the nervous system and supplements or substitutes functionality in the patient's body” Wright et al, 2016 13

Neurorobotics Robotic devices for stroke rehab Robotic limbs Wearable exoskeletons 14

Our research interests artificial system neuronal system  M ETHOD - Exploiting techniques and methodologies of engineering for biomedical applications  Understanding by building  Multi-scale experimental approach  Innovative ‘experimental’ models ( neurohybrid )  F INAL G OAL - Neurorehabilitation  Neural Interfaces (including neuroprosthetics) : interfacing neuronal circuits with artificial devices  Neuromodulation : drive neuronal dynamics  Neurorobotics: perform controlled training on patients and monitor recovery 15

Our multi-scale approach Neural Interfaces Neurorobotics Neuromodulation translational methodology 16

Our ‘macro’ scientific & technological questions  How can we interact with a neuronal system and thus modulate its dynamics?  How can we interface the neuronal element with an artificial one?  How can we restore an injured or pathological communication through an artificial device ? 17

The brain is one of the most complex system of the known Universe 10 11 neurons 10 15 synapses The brain is a world consisting of a number of unexplored continents and great stretches of unknown territory Santiago Ramón y Cajal 18

Lessons from Neuroengineering  Reduce the complexity of the system by developing a simple experimental model  Use the model to test technological solutions for brain repair 19

Multi-scale approach Human brain In vivo brain Network Neurons ‘In the future, sensory, motor, and modulatory BMIs are likely to take advantage of a continuous dialog between the nervous system and artificial computational devices. (...) Thanks to their controllability and relative simplicity, artificially embodied in vitro networks Molecules provide excellent test beds for studying plasticity mechanisms . It is not hard to imagine that this electrical training and modulation of cortical tissue could form the basis of future adaptive, closed-loop risk BMIs ’ S.M. Potter, GeorgiaTech, USA Frontiers in Neuroscience (2010 ) complexity 20

From in vitro… 21

In vitro cortical cultures Micro-Electrode Arrays (MEAs) Primary cultures of rat cortical neurons Dissection + Enzymatic digestion + Mechanical dissociation Rat embryos (E18) ~ 50.000 cells 22

Spikes and Bursts in electrophysiology  The electrophysiological signal , acquired from a single microelectrode is generally characterized by two different patterns of activity :  Spike – single over-threshold signal representing the electrical activity of one or more neurons (i.e. 1-3 cells).  Burst – sequence of highly packed spikes often occurring simultaneously on several channels and giving rise to a phenomenon generally known as ‘ network burst ’ . 23

How can we interact with a neuronal system and thus modulate its dynamics? 24

An in vitro model of neural dynamics Electrical modulation V Pasquale Pharmacological modulation BASAL BIC 30 μ M I Colombi ‘In recent years, in vitro neuronal cultures have been recognized as a successful model system of neuronal activity’ Orlandi JG et al. Nature Physics , 2013 Network patterning M Bisio Chiappalone et al. Neurocomputing , 2005; Chiappalone, et al . Brain Res , 2006; Chiappalone, et al . Int J Neural Sys , 2007; Maccione et al. J Neurosci Methods , 2009; Bologna, et al . Neural Networks , 2010; Pasquale et al. J Comput Neurosci , 2010; Bisio et al, PLoS One , 2014; 25 Kanner et al, JoVE , 2015; Pasquale et al. Scientific Reports 2017

Network bursts are typical features of in vitro neuronal cultures Spontaneous activity Evoked activity Burst 20  V 100 ms 10 ms Spike V Pasquale Pasquale, et al . J Comput Neurosci , 2010; Pasquale, et al. Scientific Reports , 2017 26

The concept of brain modularity  Brain is redundant and intrinsically modular, being composed of local networks that are embedded in networks of networks ( Meunier et al., 2009; Levy et al, 2012) Ref: Betzel et al, Nat Communications 2018 Ref: Park & Friston, Science 2013 Ref: Yamamoto et al, Science Advances 2018 28

Brain modularity in vitro: network patterning 100 s 100 s 100 ms 100 ms C1 C1 33 1 33 1 12 2 12 2 13 3 13 3 34 4 34 4 C2 62 1 C2 62 1 71 2 71 2 72 3 72 3 73 4 73 4 83 5 83 5 82 6 82 6 63 7 63 7 C3 75 1 C3 75 1 58 2 58 2 76 3 76 3 86 4 86 4 87 5 87 5 77 6 77 6 56 7 56 7 55 8 55 8 78 9 78 9 66 10 66 10 67 11 67 11 68 12 68 12 C4 17 1 C4 17 1 36 2 36 2 C5 84 1 C5 84 1 14 2 14 2 85 3 85 3 50 100 150 200 250 300 50 100 150 200 250 Time [sec] Time [sec] Bisio et al, PLoS One , 2014; Kanner et al, JoVE , 2015 29 M Bisio, in collaboration with TAU (S. Kanner, P. Bonifazi)

Multimodular systems (electrophysiology) Averna et al, submitted (LNCS) 30

Multimodular systems (electrophysiology) C1 C3 C2 Before lesion After lesion Averna et al, submitted (LNCS) 31

How can we interface the neuronal element with an artificial one? 32

The first closed-loop system  A brain with a body , i.e. a In vitro brain of a sea lamprey brain with an artificial sensory system and an artificial motor nOMI LEFT RIGHT nOMP system PRRN pulse  First example of a closed-loop spike generator detection system : an in vitro brain of a cancel DECODING artifact CODING sea lamprey bidirectionally connected to a mobile robot. from light to motor sensors actuators Karniel A, et al . J Neural Eng , 2005 Kositsky, Chiappalone et al. Front Neurorobotics , 2010 Mussa-Ivaldi FA, et al. Front Neurosci , 2010 Mobile Robot FA Mussa-Ivaldi M Kositsky V Sanguineti 33

Our neurorobotic system J Tessadori Spike Decoding detection Wheel Speed Commands Filter Amplifier Robot Tetanus Distance Sensors Readings Coding Stimulation Tessadori et al. Living Machines 2013; Tessadori and Chiappalone JoVE , 2015 34

Hybrid communication in neurorobotics experiments  u ( t ) r s ( t ) y ( t ) y ( t ) r y ˆ t ( )  ( t ) s ( t ) x ( t ) FROM ROBOT TO ROBOT CODING DECODING SPIKE SPIKE SENSORY LINEAR NEURAL FIRING RATE DETECTION GENERATION NEURAL RECEPTIVE DECODING CODE ESTIMATION AND ARTIFACT MECHANISM PREPARATION FIELDS BLANKING Target behavior : ‘ Braitenberg Left Right vehicle’ which Output Output (learns to) avoids obstacles Left Right Input Input J Tessadori Tessadori et al. Living Machines 2013; Tessadori and Chiappalone JoVE , 2015 35