Design of an experimental setup for delivering intracortical microstimulation in vivo via Spiking Neural Network.

Electroceutical approaches for the treatment of neurological disorders, such as stroke, can take advantage of neuromorphic engineering, to develop devices able to achieve a seamless interaction with the neural system. This paper illustrates the development and test of a hardware-based Spiking Neural Network (SNN) to deliver neural-like stimulation patterns in an open-loop fashion. Neurons in the SNN have been designed by following the Hodgkin-Huxley formalism, with parameters taken from neuroscientific literature. We then built the set-up to deliver the SNN-driven stimulation in vivo. We used deeply anesthetized healthy rats to test the potential effect of the SNN-driven stimulation. We analyzed the neuronal firing activity pre- and post-stimulation in both the primary somatosensory and the rostral forelimb area. Our results showed that the SNN-based neurostimulation was able increase the spontaneous level of neuronal firing at both monitored locations, as found in the literature only for closed-loop stimulation. This study represents the first step towards translating the use of neuromorphic-based devices into clinical applications.Clinical Relevance- Stroke represents one of the leading causes of long-term disability and death worldwide. Intracortical microstimulation is an effective approach for restoring lost sensory motor integration by promoting plasticity among the affected brain areas. Stimulation delivered via neuromorphic-based open-loop systems (i.e. neuromorphic prostheses) can pave the way to novel electroceutical strategies for brain repair.

NeBULA: A Standardized Protocol for the Benchmarking of Robotic-based Upper Limb Neurorehabilitation.

The use of robotic technologies in neurorehabilitation is growing, because they allow highly repeatable exercise protocols and patient-tailored therapies. However, there is a lack of objective methods for assessing these technologies, which makes it difficult to determine their value in rehabilitation settings. While there exist many outcome measurements for motor assessment from a clinical standpoint (such as the Fugl-Meyer scale), the evaluation of performance and clinical benefits of technology for rehabilitation still lacks a standardized approach from a technical standpoint.In this work, we describe NeBULA (Neuromechanical Biomarkers for Upper Limb Assessment), a benchmarking platform for evaluating robotic technology for upper limb neurorehabilitation. By utilizing standardized neuromechanical biomarkers, NeBULA aims at providing a groundwork for assessing and comparing neurorehabilitation robots. We describe its implementation and preliminary results assessing a novel upper limb exoskeleton.Clinical Relevance- Standardized evaluation of neurorehabilitation robots can lead to better patient outcomes, optimizing resources by identifying the most effective technology and by boosting their use in clinical practice. This would provide quantitative and objective information to complement clinical motor evaluation - preventing suboptimal treatments and ensuring that patients receive personalized care. It can also facilitate the transfer of technologyto clinics, identifying the most promising ones for further investment and research.

Model-based online implementation of spike detection algorithms for neuroengineering applications.

Traditional methods for the development of a neuroprosthesis to perform closed-loop stimulation can be complex and the necessary technical knowledge and experience often present a high barrier for adoption. This paper takes a novel Model-Based Design approach to simplifying such closed-loop system development, and thereby lowering the adoption barrier. This work implements a computational model of different spike detection algorithms in Simulink® and compares their performances by taking advantage of synthetic neural signals to evaluate suitability for the intended embedded implementation. Clinical Relevance--- Closed-loop systems have been demonstrated to be suitable for brain repair strategies. Coupling two different brain areas by means of a neuroprosthesis can potentially lead to restoration of communication by inducing activity-dependent plasticity.

Extracellular recordings from locally dense microelectrode arrays coupled to dissociated cortical cultures.

High-density microelectrode arrays (MEAs) enabled by recent developments of microelectronic circuits (CMOS-MEA) and providing spatial resolutions down to the cellular level open the perspective to access simultaneously local and overall neuronal network activities expressed by in vitro preparations. The short inter-electrode separation results in a gain of information on the micro-circuit neuronal dynamics and signal propagation, but requires the careful evaluation of the time resolution as well as the assessment of possible cross-talk artifacts. In this respect, we have realized and tested Pt high-density (HD)-MEAs featuring four local areas with 10microm inter-electrode spacing and providing a suitable noise level for the assessment of the high-density approach. First, simulated results show how possible artifacts (duplicated spikes) can be theoretically observed on nearby microelectrodes only for very high-shunt resistance values (e.g. R(sh)=50 kOmega generates up to 60% of false positives). This limiting condition is not compatible with typical experimental conditions (i.e. dense but not confluent cultures). Experiments performed on spontaneously active cortical neuronal networks show that spike synchronicity decreases by increasing the time resolution and analysis results show that the detected synchronous spikes on nearby electrodes are likely to be unresolved (in time) fast local propagations. Finally, functional connectivity analysis results show stronger local connections than long connections spread homogeneously over the whole network demonstrating the expected gain in detail provided by the spatial resolution.

Self-organization and neuronal avalanches in networks of dissociated cortical neurons.

Dissociated cortical neurons from rat embryos cultured onto micro-electrode arrays exhibit characteristic patterns of electrophysiological activity, ranging from isolated spikes in the first days of development to highly synchronized bursts after 3-4 weeks in vitro. In this work we analyzed these features by considering the approach proposed by the self-organized criticality theory: we found that networks of dissociated cortical neurons also generate spontaneous events of spreading activity, previously observed in cortical slices, in the form of neuronal avalanches. Choosing an appropriate time scale of observation to detect such neuronal avalanches, we studied the dynamics by considering the spontaneous activity during acute recordings in mature cultures and following the development of the network. We observed different behaviors, i.e. sub-critical, critical or super-critical distributions of avalanche sizes and durations, depending on both the age and the development of cultures. In order to clarify this variability, neuronal avalanches were correlated with other statistical parameters describing the global activity of the network. Criticality was found in correspondence to medium synchronization among bursts and high ratio between bursting and spiking activity. Then, the action of specific drugs affecting global bursting dynamics (i.e. acetylcholine and bicuculline) was investigated to confirm the correlation between criticality and regulated balance between synchronization and variability in the bursting activity. Finally, a computational model of neuronal network was developed in order to interpret the experimental results and understand which parameters (e.g. connectivity, excitability) influence the distribution of avalanches. In summary, cortical neurons preserve their capability to self-organize in an effective network even when dissociated and cultured in vitro. The distribution of avalanche features seems to be critical in those cultures displaying medium synchronization among bursts and poor random spiking activity, as confirmed by chemical manipulation experiments and modeling studies.

Neuroengineering Tools For Studying The Effect Of Intracortical Microstimulation In Rodent Models.

Intracortical microstimulation can be successfully used to manipulate neuronal activity and connectivity, thus representing a potentially powerful tool to steer neuroplasticity occurring after brain injury. Activity-dependent stimulation (ADS), in which the spikes recorded from a single neuron are used to trigger stimulation at another cortical location, is able to potentiate cortical connections between distant brain areas. Here, we developed an experimental procedure and a computational pipeline aimed at investigating the ability of ADS to induce changes in intra-cortical activity of healthy anesthetized rats.

In vitro cortical neuronal networks as a new high-sensitive system for biosensing applications.

By taking advantages of the main features of the microelectrode array (MEA) technology (i.e. multisite recordings, stable and long-term coupling with the biological preparation), we analyzed the changes in activity patterns induced by applying specific substances to dissociated cortical neurons from rat-embryos (E18). Data were recorded simultaneously from 60 electrodes, and the electrophysiological behavior was investigated during the third week in vitro, both at the spike and burst level. The analysis of the electrophysiological activity modulation, by applying agonists of the ionotropic glutamate receptors at low (i.e. 0.2-1-5 microM) and high (i.e. 50-100 microM) concentrations, is presented. Preliminary results show that the dynamics of the in vitro cortical neurons is very sensitive to pharmacological manipulation of the glutamatergic transmission and the effects on the network behavior are strictly dependent from the drug concentration. In particular, the addition of a high-dose of agonist determined a global and irreversible depression of the network activity, while, in the low-concentration case, the electrophysiological behavior showed different results, depending on the type of receptor involved. From these observations, we are encouraged to think of a more engineered system, based on in vitro cortical neurons, as a novel sensitive system for drug (pre)-screening and neuropharmacological evaluations.

Networks of neurons coupled to microelectrode arrays: a neuronal sensory system for pharmacological applications.

Two main features make microelectrode arrays (MEAs) a valuable tool for electrophysiological measurements under the perspective of pharmacological applications, namely: (i) they are non-invasive and permit, under appropriate conditions, to monitor the electrophysiological activity of neurons for a long period of time (i.e. from several hours up to months); (ii) they allow a multi-site recording (up to tens of channels). Thus, they should allow a high-throughput screening while reducing the need for animal experiments. In this paper, by taking advantages of these features, we analyze the changes in activity pattern induced by the treatment with specific substances, applied on dissociated neurons coming from the chick-embryo spinal cord. Following pioneering works by Gross and co-workers (see e.g. Gross and Kowalski, 1991. Neural Networks, Concepts, Application and Implementation, vol. 4. Prentice Hall, NJ, pp. 47-110; Gross et al., 1992. Sensors Actuators, 6, 1-8.), in this paper analysis of the drugs' effects (e.g. NBQX, CTZ, MK801) to the collective electrophysiological behavior of the neuronal network in terms of burst activity, will be presented. Data are simultaneously recorded from eight electrodes and besides variations induced by the drugs also the correlation between different channels (i.e. different area in the neural network) with respect to the chemical stimuli will be introduced (Bove et al., 1997. IEEE Trans. Biomed. Eng., 44, 964-977.). Cultured spinal neurons from the chick embryo were chosen as a neurobiological system for their relative simplicity and for their reproducible spontaneous electrophysiological behavior. It is well known that neuronal networks in the developing spinal cord are spontaneously active and that the presence of a significant and reproducible bursting activity is essential for the proper formation of muscles and joints (Chub and O'Donovan, 1998. J. Neurosci., 1, 294-306.). This fact, beside a natural variability among different biological preparations, allows a comparison also among different experimental session giving reliable results and envisaging a definition of a bioelectronic 'neuronal sensory system'.

Separating burst from background spikes in multichannel neuronal recordings using return map analysis.

We propose a preprocessing method to separate coherent neuronal network activity, referred to as "bursts", from background spikes. High background activity in neuronal recordings reduces the effectiveness of currently available burst detection methods. For long-term, stationary recordings, burst and background spikes have a bimodal ISI distribution which makes it easy to select the threshold to separate burst and background spikes. Finite, nonstationary recordings lead to noisy ISIs for which the bimodality is not that clear. We introduce a preprocessing method to separate burst from background spikes to improve burst detection reliability because it efficiently uses both single and multichannel activity. The method is tested using a stochastic model constrained by data available in the literature and recordings from primary cortical neurons cultured on multielectrode arrays. The separation between burst and background spikes is obtained using the interspike interval return map. The cutoff threshold is the key parameter to separate the burst and background spikes. We compare two methods for selecting the threshold. The 2-step method, in which threshold selection is based on fixed heuristics. The iterative method, in which the optimal cutoff threshold is directly estimated from the data. The proposed preprocessing method significantly increases the reliability of several established burst detection algorithms, both for simulated and real recordings. The preprocessing method makes it possible to study the effects of diseases or pharmacological manipulations, because it can deal efficiently with nonstationarity in the data.

Nickel modulates the electrical activity of cultured cortical neurons through a specific effect on N-methyl-D-aspartate receptor channels.

Nickel (Ni(2+)) is a toxic metal that affects the function of several ionic channels. In the N-methyl-d-aspartate (NMDA) subtype of glutamate receptor (NR), it causes activity enhancement of the channels containing the NR2B subunit and voltage-independent inhibition of those containing NR2A. Thus, it may represent a functional marker for the identification of NR native channel subunits. We investigated the effect of Ni(2+) on spontaneous NR currents in cortical neurons, dissociated from 18-day rat embryos and maintained in culture for up to ∼40 days. In whole-cell voltage-clamp at -60 mV, in a Mg(2+)-free bath containing the alpha-amino-3-hydroxy-5-methyl-4-isoxazole propionate (AMPA) antagonist 2,3-dihydroxy-6-nitro-7-sulfamoyl-benzo[f]quinoxaline-2,3-dione (NBQX) (10 µM), spontaneous currents were blocked by 10 µM D(-)-2-Amino-5-phosphonopentanoic acid (APV) (10 µM), and by NR2B antagonists, ifenprodil (10 µM) or Ro25-6981 (Ro25, 1 µM), indicating that they are due to NRs containing predominantly the NR2B subunit. In the presence of Ni(2+) (30 µM) the amplitude and the frequency of spontaneous currents were increased and the decay time decreased. A higher dose (300 µM) blocked all electrical activity. In current-clamp, Ni(2+) (30 µM) caused a ∼5 mV reversible depolarization. The effect of Ni(2+), as well as that of NR2B antagonists, was almost independent of days in vitro (DIV) in the range from 18 to 33 DIV. The electrical activity of the neuronal networks measured by microelectrode arrays (MEAs) was also affected by Ni(2+), which caused a decrease in firing rate, but an increase in burst duration, while Ro25 (1-10 µM) caused a decrease in both firing rate and burst duration. Finally, reverse transcription polymerase chain reaction (RT-PCR) revealed a predominant expression of NR2B, with no modification during DIV. These results demonstrate that, in these cultured cells, the NR spontaneous current is almost entirely due by NR2B-containing receptors and that Ni(2+) affects the electrical activity through a specific effect on NR channels.

Development of micro-electrode array based tests for neurotoxicity: assessment of interlaboratory reproducibility with neuroactive chemicals.

Neuronal assemblies within the nervous system produce electrical activity that can be recorded in terms of action potential patterns. Such patterns provide a sensitive endpoint to detect effects of a variety of chemical and physical perturbations. They are a function of synaptic changes and do not necessarily involve structural alterations. In vitro neuronal networks (NNs) grown on micro-electrode arrays (MEAs) respond to neuroactive substances as well as the in vivo brain. As such, they constitute a valuable tool for investigating changes in the electrophysiological activity of the neurons in response to chemical exposures. However, the reproducibility of NN responses to chemical exposure has not been systematically documented. To this purpose six independent laboratories (in Europe and in USA) evaluated the response to the same pharmacological compounds (Fluoxetine, Muscimol, and Verapamil) in primary neuronal cultures. Common standardization principles and acceptance criteria for the quality of the cultures have been established to compare the obtained results. These studies involved more than 100 experiments before the final conclusions have been drawn that MEA technology has a potential for standard in vitro neurotoxicity/neuropharmacology evaluation. The obtained results show good intra- and inter-laboratory reproducibility of the responses. The consistent inhibitory effects of the compounds were observed in all the laboratories with the 50% Inhibiting Concentrations (IC(50)s) ranging from: (mean ± SEM, in µM) 1.53 ± 0.17 to 5.4 ± 0.7 (n = 35) for Fluoxetine, 0.16 ± 0.03 to 0.38 ± 0.16 µM (n = 35) for Muscimol, and 2.68 ± 0.32 to 5.23 ± 1.7 (n = 32) for Verapamil. The outcome of this study indicates that the MEA approach is a robust tool leading to reproducible results. The future direction will be to extend the set of testing compounds and to propose the MEA approach as a standard screen for identification and prioritization of chemicals with neurotoxicity potential.

Low-frequency stimulation enhances burst activity in cortical cultures during development.

The intact brain is continuously targeted by a wealth of stimuli with distinct spatio-temporal patterns which modify, since the very beginning of development, the activity and the connectivity of neuronal networks. In this paper, we used dissociated neuronal cultures coupled to microelectrode arrays (MEAs) to study the response of cortical neuron assemblies to low-frequency stimuli constantly delivered over weeks in vitro. We monitored the spontaneous activity of the cultures before and after the stimulation sessions, as well as their evoked response to the stimulus. During in vitro development, the vast majority of the cultures responded to the stimulation by significantly increasing the bursting activity and a widespread stabilization of electrical activity was observed after the third week of age. A similar trend was present between the spontaneous activity of the networks observed over 30 min after the stimulus and the responses evoked by the stimulus itself, although no significant differences in spontaneous activity were detected between stimulated and non-stimulated cultures belonging to the same preparations. The data indicate that the stimulation had a delayed effect modulating responsiveness capability of the network without directly affecting its intrinsic in vitro development.

Connecting neurons to a mobile robot: an in vitro bidirectional neural interface.

One of the key properties of intelligent behaviors is the capability to learn and adapt to changing environmental conditions. These features are the result of the continuous and intense interaction of the brain with the external world, mediated by the body. For this reason "embodiment" represents an innovative and very suitable experimental paradigm when studying the neural processes underlying learning new behaviors and adapting to unpredicted situations. To this purpose, we developed a novel bidirectional neural interface. We interconnected in vitro neurons, extracted from rat embryos and plated on a microelectrode array (MEA), to external devices, thus allowing real-time closed-loop interaction. The novelty of this experimental approach entails the necessity to explore different computational schemes and experimental hypotheses. In this paper, we present an open, scalable architecture, which allows fast prototyping of different modules and where coding and decoding schemes and different experimental configurations can be tested. This hybrid system can be used for studying the computational properties and information coding in biological neuronal networks with far-reaching implications for the future development of advanced neuroprostheses.

Computationally inexpensive methods for intra-cardiac atrial bipolar electrogram compression.

AIM: This paper reports studies of mathematical algorithms for intra-cardiac atrial bipolar electrogram compression suitable with implementation on implantable devices. PATIENTS AND METHODS: Bipolar intra-cardiac electrograms (IEGMs) of high right atrium were obtained from 20 patients who underwent electrophysiological studies for arrhythmias. Four thousand seven hundred and eighty-two seconds of IEGM were collected and divided into three rhythm groups: sinus rhythm (SR), atrial fibrillation (AF) and atrial flutter (AFL). Since mathematical algorithms suitable for use with implantable devices demand low computational cost, we employed piecemeal linear approximation methods (ZOP--Zero Order Prediction and SAPA--Scan Along Polygonal Approximation), and beat detection method (Peak) both or which need small numbers of operations to perform electrogram compression. Compression ratio (CR) and percent root mean square difference (PRD) were used to compare the three methods, with statistical analyses performed using paired t-test. RESULTS AND CONCLUSION: The best performance was obtained using the Peak method which reaches an average CR of 10.6 in the case of SR group, 2.8 for AF, and 3.6 for AFL groups, respectively, while PRD lies below 2% for SR and AFL groups and 6% for the AF group. Results show that, for bipolar electrograms, the Peak method reaches statistically significant better performance (P<0.001) in all cases except for Peak vs SAPA applied to AF (P=0.2). The number of operations necessary to compress the data indicate that time consumption can be reduced to be suitable for real time compression in implantable devices. The Peak method, which was assumed to receive the instant of occurrence of each recognized beat, from the hardware of the device, requires fewer operations than ZOP and SAPA. Increasing the length of electrograms recorded in pacemakers will enhance the amount of information provided by the implantable device, allowing more detailed characterization of the intra-cardiac activity and leading to new perspectives in arrhythmia diagnosis and therapy.