Espen Hagen

Towards reliable spike-train recordings from thousands of neurons with multielectrodes.

The new generation of silicon-based multielectrodes comprising hundreds or more electrode contacts offers unprecedented possibilities for simultaneous recordings of spike trains from thousands of neurons. Such data will not only be invaluable for finding out how neural networks in the brain work, but will likely be important also for neural prosthesis applications. This opportunity can only be realized if efficient, accurate and validated methods for automatic spike sorting are provided. In this review we describe some of the challenges that must be met to achieve this goal, and in particular argue for the critical need of realistic model data to be used as ground truth in the validation of spike-sorting algorithms.

LFPy: a tool for biophysical simulation of extracellular potentials generated by detailed model neurons

Electrical extracellular recordings, i.e., recordings of the electrical potentials in the extracellular medium between cells, have been a main work-horse in electrophysiology for almost a century. The high-frequency part of the signal (≳500 Hz), i.e., the multi-unit activity (MUA), contains information about the firing of action potentials in surrounding neurons, while the low-frequency part, the local field potential (LFP), contains information about how these neurons integrate synaptic inputs. As the recorded extracellular signals arise from multiple neural processes, their interpretation is typically ambiguous and difficult. Fortunately, a precise biophysical modeling scheme linking activity at the cellular level and the recorded signal has been established: the extracellular potential can be calculated as a weighted sum of all transmembrane currents in all cells located in the vicinity of the electrode. This computational scheme can considerably aid the modeling and analysis of MUA and LFP signals. Here, we describe LFPy, an open source Python package for numerical simulations of extracellular potentials. LFPy consists of a set of easy-to-use classes for defining cells, synapses and recording electrodes as Python objects, implementing this biophysical modeling scheme. It runs on top of the widely used NEURON simulation environment, which allows for flexible usage of both new and existing cell models. Further, calculation of extracellular potentials using the line-source-method is efficiently implemented. We describe the theoretical framework underlying the extracellular potential calculations and illustrate by examples how LFPy can be used both for simulating LFPs, i.e., synaptic contributions from single cells as well a populations of cells, and MUAs, i.e., extracellular signatures of action potentials.

ViSAPy: a Python tool for biophysics-based generation of virtual spiking activity for evaluation of spike-sorting algorithms.

BACKGROUND New, silicon-based multielectrodes comprising hundreds or more electrode contacts offer the possibility to record spike trains from thousands of neurons simultaneously. This potential cannot be realized unless accurate, reliable automated methods for spike sorting are developed, in turn requiring benchmarking data sets with known ground-truth spike times. NEW METHOD We here present a general simulation tool for computing benchmarking data for evaluation of spike-sorting algorithms entitled ViSAPy (Virtual Spiking Activity in Python). The tool is based on a well-established biophysical forward-modeling scheme and is implemented as a Python package built on top of the neuronal simulator NEURON and the Python tool LFPy. RESULTS ViSAPy allows for arbitrary combinations of multicompartmental neuron models and geometries of recording multielectrodes. Three example benchmarking data sets are generated, i.e., tetrode and polytrode data mimicking in vivo cortical recordings and microelectrode array (MEA) recordings of in vitro activity in salamander retinas. The synthesized example benchmarking data mimics salient features of typical experimental recordings, for example, spike waveforms depending on interspike interval. COMPARISON WITH EXISTING METHODS ViSAPy goes beyond existing methods as it includes biologically realistic model noise, synaptic activation by recurrent spiking networks, finite-sized electrode contacts, and allows for inhomogeneous electrical conductivities. ViSAPy is optimized to allow for generation of long time series of benchmarking data, spanning minutes of biological time, by parallel execution on multi-core computers. CONCLUSION ViSAPy is an open-ended tool as it can be generalized to produce benchmarking data or arbitrary recording-electrode geometries and with various levels of complexity.

Hybrid Scheme for Modeling Local Field Potentials from Point-Neuron Networks

With rapidly advancing multi-electrode recording technology, the local field potential (LFP) has again become a popular measure of neuronal activity in both research and clinical applications. Proper understanding of the LFP requires detailed mathematical modeling incorporating the anatomical and electrophysiological features of neurons near the recording electrode, as well as synaptic inputs from the entire network. Here we propose a hybrid modeling scheme combining efficient point-neuron network models with biophysical principles underlying LFP generation by real neurons. The LFP predictions rely on populations of network-equivalent multicompartment neuron models with layer-specific synaptic connectivity, can be used with an arbitrary number of point-neuron network populations, and allows for a full separation of simulated network dynamics and LFPs. We apply the scheme to a full-scale cortical network model for a ∼1 mm2 patch of primary visual cortex, predict laminar LFPs for different network states, assess the relative LFP contribution from different laminar populations, and investigate effects of input correlations and neuron density on the LFP. The generic nature of the hybrid scheme and its public implementation in hybridLFPy form the basis for LFP predictions from other and larger point-neuron network models, as well as extensions of the current application with additional biological detail.

Focal local field potential signature of the single-axon monosynaptic thalamocortical connection

A resurgence has taken place in recent years in the use of the extracellularly recorded local field potential (LFP) to investigate neural network activity. To probe monosynaptic thalamic activation of cortical postsynaptic target cells, so called spike-trigger-averaged LFP (stLFP) signatures have been measured. In these experiments, the cortical LFP is measured by multielectrodes covering several cortical lamina and averaged on spontaneous spikes of thalamocortical (TC) cells. Using a well established forward-modeling scheme, we investigated the biophysical origin of this stLFP signature with simultaneous synaptic activation of cortical layer-4 neurons, mimicking the effect of a single afferent spike from a single TC neuron. Constrained by previously measured intracellular responses of the main postsynaptic target cell types and with biologically plausible assumptions regarding the spatial distribution of thalamic synaptic inputs into layer 4, the model predicted characteristic contributions to monosynaptic stLFP signatures both for the regular-spiking (RS) excitatory neurons and the fast-spiking (FS) inhibitory interneurons. In particular, the FS cells generated stLFP signatures of shorter temporal duration than the RS cells. Added together, a sum of the stLFP signatures of these two principal synaptic targets of TC cells were observed to resemble experimentally measured stLFP signatures. Outside the volume targeted by TC afferents, the resulting postsynaptic LFP signals were found to be sharply attenuated. This implies that such stLFP signatures provide a very local measure of TC synaptic activation, and that newly developed inverse current-source density (CSD)-estimation methods are needed for precise assessment of the underlying spatiotemporal CSD profiles. SIGNIFICANCE STATEMENT Despite its long history and prevalent use, the proper interpretation of the extracellularly recorded local field potential (LFP) is still not fully established. Here we investigate by biophysical modeling the origin of the focal LFP signature of the single-axon monosynaptic thalamocortical connection as measured by spike-trigger-averaging of cortical LFPs on spontaneous spikes of thalamocortical neurons. We find that this LFP signature is well accounted for by a model assuming thalamic projections to two cortical layer-4 cell populations: one excitatory (putatively regular-spiking cells) and one inhibitory (putatively fast-spiking cells). The LFP signature is observed to decay sharply outside the cortical region receiving the thalamocortical projection, implying that it indeed provides a very local measure of thalamocortical synaptic activation.

Biophysical Network Modelling of the dLGN Circuit: Different Effects of Triadic and Axonal Inhibition on Visual Responses of Relay Cells

Despite its prominent placement between the retina and primary visual cortex in the early visual pathway, the role of the dorsal lateral geniculate nucleus (dLGN) in molding and regulating the visual signals entering the brain is still poorly understood. A striking feature of the dLGN circuit is that relay cells (RCs) and interneurons (INs) form so-called triadic synapses, where an IN dendritic terminal can be simultaneously postsynaptic to a retinal ganglion cell (GC) input and presynaptic to an RC dendrite, allowing for so-called triadic inhibition. Taking advantage of a recently developed biophysically detailed multicompartmental model for an IN, we here investigate putative effects of these different inhibitory actions of INs, i.e., triadic inhibition and standard axonal inhibition, on the response properties of RCs. We compute and investigate so-called area-response curves, that is, trial-averaged visual spike responses vs. spot size, for circular flashing spots in a network of RCs and INs. The model parameters are grossly tuned to give results in qualitative accordance with previous in vivo data of responses to such stimuli for cat GCs and RCs. We particularly investigate how the model ingredients affect salient response properties such as the receptive-field center size of RCs and INs, maximal responses and center-surround antagonisms. For example, while triadic inhibition not involving firing of IN action potentials was found to provide only a non-linear gain control of the conversion of input spikes to output spikes by RCs, axonal inhibition was in contrast found to substantially affect the receptive-field center size: the larger the inhibition, the more the RC center size shrinks compared to the GC providing the feedforward excitation. Thus, a possible role of the different inhibitory actions from INs to RCs in the dLGN circuit is to provide separate mechanisms for overall gain control (direct triadic inhibition) and regulation of spatial resolution (axonal inhibition) of visual signals sent to cortex.

An automated online positioning system and simulation environment for multi-electrodes in extracellular recordings

Extracellular recordings are a key tool to record the activity of neurons in vivo. Especially in the case of experiments with behaving animals, however, the tedious procedure of electrode placement can take a considerable amount of expensive and restricted experimental time. Furthermore, due to tissue drifts and other sources of variability in the recording setup, the position of the electrodes with respect to the recorded neurons can change causing low recording quality. The contributions of this work are threefold. We introduce a quality measure for the recording position of the electrode which should be maximized during recordings and is especially suitable for the use of multi-electrodes. An automated positioning system based on this quality measure is proposed. The system is able to find favorable recording positions and adapts the electrode position smoothly to changes of the neuron positions. Finally, we evaluate the system using a new simulator for extracellular recordings based on realistically reconstructed 3D neurons.

Hybrid scheme for modeling local field potentials from point-neuron networks

Measurement of the local field potential (LFP) has become routine for assessment of neuronal activity in neuroscientific and clinical applications, but its interpretation remains nontrivial. Understanding the LFP requires accounting for both anatomical and electrophysiological features of neurons near the recording electrode as well as the entire large-scale neuronal circuitry generating synaptic input to these cells. The direct simulation of LFPs in biophysically detailed network models is computationally daunting. Here, we instead propose a hybrid modeling scheme combining the efficiency of simplified point-neuron network models (Fig. ​(Fig.1A)1A) with the biophysical principles underlying LFP generation by multicompartment neurons [1] (Fig ​(Fig1C).1C). We apply this scheme to a model representing a full-scale cortical network under about 1 square millimeter surface of cat primary visual cortex [2] (Fig. 1A,B) with layer-specific connectivity [3] to predict laminar LFPs (Fig. ​(Fig.1D)1D) for different network states, assess the relative contribution of local neuron populations to the LFP, investigate the role of input correlations and neuron density, and validate linear LFP predictions based on population firing rates. The hybrid scheme is accompanied by our open-source software, hybridLFPy (github.com/espenhgn/hybridLFPy). Figure 1 Overview of the hybrid scheme for modeling LFP generated by a cortical network model. A Sketch of point-neuron network model [1]. B Spikes of point neurons in the network for spontaneous and evoked activity. C Populations of multi-compartment neurons ...

Hybrid scheme for modeling LFPs from spiking cortical network models

While recordings of extracellular potentials (EP) remain a common method for experimentally measuring neural activity, the interpretation of the low-frequency part, the local field potential (LFP), is not straightforward. Cortical LFPs seem to mainly stem from synaptic inputs, but the net LFP signal from several contributing laminar populations is difficult to assess, as the individual contributions will depend on their locations, the morphologies of the postsynaptic neurons, the spatial distribution of active synapses, and the level of correlations in synaptic inputs [1]. While most comprehensive cortical-network simulations are done with single-compartment models [2], multicompartmental neuronal modeling is in general required to calculate LFPs [1]. Here we present a hybrid LFP modeling approach where a network of single-compartment LIF neurons generates the spiking activity (Figure ​(Figure1A),1A), while detailed multicompartment neuronal models are used to calculate the accompanying LFP (Figure 1B-C). Our model describes a 1mm2 patch of cat V1, and we incorporate spatially specific pre- to post-synaptic inter- and intra-layer connectivity constrained by experimental observations [3] using reconstructed neuron morphologies of excitatory and inhibitory neurons in layers L2/3-L6 with passive membrane properties. Model specifications of neuron and synapse numbers within populations are taken from [2], while spatial connectivity profiles are based on [3]. Our hybrid simulation framework allows detailed analysis of how the LFP correlates with network activity and connectivity, and how spatially specific synapse distributions influence the LFP. Spiking network simulations [2] were implemented in NEST (http://www.nest-initiative.org), while simulations of LFPs from morphologically realistic neurons used LFPy (compneuro.umb.no/LFPy) along with NEURON [4]. Figure 1 Schematic illustration of the hybrid scheme. (A) Spiking activity generated in network simulations using single-compartment neurons [2] are used as input to multicompartmental neuron models to generate LFPs (B). LFP contributions from each postsynaptic ...

Modeling Extracellular Potentials in Microelectrode Array Recordings

Microelectrode Array (MEA) measurements from in vitro slices has become an important research tool in neuroscience, however the interpretation of such recordings is not always straightforward. We have developed a modeling framework for emulating in vitro MEA recordings that takes into account both the measurement physics of the MEA set-up, and the underlying neural activity of the slice, resulting in simulated data that closely resembles experimental recordings. Our modeling framework may aid interpretation of experimental data by reproducing the experimental procedure in silico, make experimentally testable predictions, and produce test-data for validating various analysis methods such as CSD estimates and spike-sorting algorithms. Our simulations are separated into two domains; the first step is simulations of neuronal activity in populations of multi-compartment model neurons, and secondly solving the electrostatic forward problem in the extracellular space. For the neuronal simulations we employ LFPy [1], a Python module built upon NEURONs Python interface [2] to obtain the transmembrane currents in every compartment of the model neurons. Then the Finite Element Method (FEM) is used to solve the Poisson equation from electrostatics and calculate the extracellular potentials in the 3D volume including the electrode sites, and test various approximation schemes. Hence, the effects of the electrodes can be assessed together with the impact of inhomogeneities and anisotropies of the extracellular medium in recordings. The approach is in principle applicable to any multicompartment neuron model (from e.g. ModelDB [3]), any neuron number or any MEA electrode set-up. We will present our modeling framework, together with an investigation of the electrode effects on the measured signals. Then we will go on to present two different applications. Firstly, we have produced spike-sorting test-data to benchmark automated spike-sorting algorithms [4] used on MEA recordings. This project is part of an international coordinated effort where such test-data will be collected and made available at http://spike.g-node.org, allowing exchange of synthetic and experimental test-data with known underlying activity, and systematic benchmarking and comparison of spike-sorting algorithms applied to such data [5]. Secondly we will present a project where we have been studying the LFP signature of single neurons receiving varying, sub-threshold sinusoidal current input measured by MEAs in an acute brain slice setting [6]. The model output is compared to corresponding experimental data, which includes the detailed reconstruction of the excited neuron.