CNS Links:
Link to the Conference Program:
CNS Program
Link to the Workshop Sched page:
Sched
Modeling extracellular potentials:
principles, methods, and applications
Workshop of the 34th Annual Computational Neuroscience Meeting - CNS 2025
09 July 2025, 9:00–17:30 (CEST), Room 5, Florence, Italy
Simulating large-scale neural activity is essential for understanding brain dynamics and linking in silico models to experimentally measurable signals like LFP, EEG, and MEG. These simulations, ranging from detailed biophysical models to simplified proxies, bridge microscale neural dynamics with meso- and macro-scale recordings, offering powerful tools to interpret data, refine analyses, and explore brain function.
Recent advances have demonstrated the clinical and theoretical value of such models, shedding light on oscillations, excitation-inhibition balance, and biomarkers of neurological disorders like epilepsy, Alzheimer's, and Parkinson’s disease.
This workshop will cover the latest methodologies, hybrid modeling approaches, and applications of brain signal simulations.
By gathering experts across disciplines, it aims to foster collaboration and advance our understanding of brain function and dysfunction.
9:15 - 09:50
How structural connectivity directs cortical traveling waves and shapes frequency gradients
Dominik Peter Koller
Brain Simulation Section, Berlin Institute of Health at Charité – Universitätsmedizin Berlin
Abstract
Waves of neural activity propagate throughout the cortex across different spectral and spatial scales. Their proposed functions range from perception modulation and memory processing to motor planning. While cortical traveling waves follow specific directions, the mechanisms directing them remain unclear. Some studies suggest a connection between cortical traveling wave directions and frequency gradients but their precise relationship is unknown. We hypothesize that structural connection strength gradients influence both cortical wave direction and frequency gradients. Our analysis reveals that the human connectome exhibits large-scale connection strength gradients. In cortical network models informed by this connectome, traveling waves emerge naturally and propagate from regions of weaker to stronger connectivity. Simultaneously, cortical frequency gradients are generated and align with these structural gradients in our simulations. We fitted our model to MEG resting-state functional connectivity data and found that the best-fitting models generate directed traveling waves and frequency gradients, supporting the role of structural connectivity in shaping cortical dynamics.
9:50 - 10:25
Modeling electric brain signals and stimulation
Gaute Einevoll
Norwegian University of Life Sciences & University of Oslo
Abstract
The combined use of volume-conductor theory and compartmental modeling of neuronal dynamics allows for biophysical computation of extracellular electric potentials such as extracellular spikes, multiunit activity (MUA), local field potentials (LFP), ECoG and EEG signals, as well as magnetic signals such as MEG (Halnes et al, Electric Brain Signals, Cambridge University Press, 2024). In the presentation this computational scheme will be presented together with example results from its use in computing brain signals generated by neurons and neuronal populations. The use of the same scheme, aided by the Reciprocity Theorem, to simulate effects of electric stimulation of neurons, will also be presented.
11:00 - 11:35
Large-scale modeling of mesoscopic networks at single-neuron resolution
Johanna Senk
Sussex AI, School of Engineering and Informatics, University of Sussex, Brighton, United Kingdom & Institute for Advanced Simulation (IAS-6), Jülich Research Centre, Jülich, Germany
Abstract
We develop and investigate spatially structured multi-layer point-neuron network models of about 3.2 million neurons and 7.7 billion synapses (Senk et al., Cerebral Cortex, 2024), covering 4x4 mm² of cortical surface at a realistic density of neurons and connections. From the simulated spiking activity, we compute biophysics-based predictions of local field potentials. The modeling framework enables a direct comparison between simulation results and extracellular recordings obtained with multi-electrode arrays.
11:35 - 12:10
Inverse Modelling of Field Potentials from Simulations of Spiking Network Models: Applications in Neuroscience Research and Clinical Settings
Pablo Martínez Cañada
Research Centre for Information and Communications Technologies (CITIC), University of Granada, Granada, Spain
Department of Computer Engineering, Automation and Robotics, University of Granada, Granada, Spain
Abstract
Network models composed of leaky integrate-and-fire (LIF) spiking neurons have been employed to address fundamental questions about neural dynamics, cortical oscillations, sensory processing, working memory, and behaviour. A key drawback of LIF neuron models is their representation as a single compartment, which prevents them from capturing the spatial distribution of transmembrane currents necessary for generating extracellular potentials. However, recent advancements have enabled an accurate biophysical approximation of extracellular signals by leveraging variables directly measured from LIF network simulations. By incorporating appropriate forward head models, LIF network models can generate realistic representations of large-scale field potentials, such as electroencephalogram (EEG) and magnetoencephalography (MEG) signals, paving the way for their application in clinical settings. Despite these advancements, our understanding of the precise mechanistic relationship between field potential recordings and key aspects of neural activity remains incomplete. In particular, it remains unclear which specific features of electrophysiological population activity (i.e., putative biomarkers) best reflect properties of the underlying microcircuit configuration. To aid this effort, our lab has developed inverse models that combine LIF network simulations, realistic extracellular signal computation, and machine learning techniques to automatically infer the most probable cortical circuit configuration that matches field potential features. Our tools serve as a benchmarking resource for model-driven interpretation of electrophysiological data and for evaluating candidate field potential biomarkers that may reflect changes in neural circuit parameters. Using mouse LFP data and human EEG recordings, we demonstrate the potential of our approach to uncover imbalances in neural circuit parameters during brain development and in Alzheimer's disease.
12:10 - 12:40
From microcircuits to mesoscopic signals: a kernel approach to efficient and interpretable LFP estimation
Nicolò Meneghetti
Sant'Anna School of Advanced Studies, Pisa, Italy
Abstract
Simulations of large-scale neural activity are increasingly used to explore the biophysical origins of brain signals. However, estimating measurable signals like the local field potential (LFP) often requires detailed multicompartmental models, which can be computationally demanding. In this talk, I will present a kernel-based method that enables efficient and accurate LFP estimation from a biologically realistic large-scale model of mouse primary visual cortex (V1). This method not only reduces computational cost but also improves interpretability by separating the contributions of specific neuronal populations. Applied to the V1 model, it reveals that LFP signals are primarily shaped by external synaptic inputs originating outside the cortex, while local V1 activity plays only a minor role. These findings highlight the kernel method as a robust tool for large-scale simulations, with strong potential for uncovering the circuit mechanisms underlying population signals.
14:15 - 14:50
Connecting biophysical models to empirical power spectra using Filtered Point Processes
Emily Patricia Stephen
Boston University (Department of Mathematics and Statistics, Center for Systems Neuroscience, Neurophotonics Center)
Abstract
Large-scale neuronal circuit simulations have been very successful in modeling extracellular potentials, but their relationship to empirical data is typically qualitative. A statistical connection to data would be valuable, for example to evaluate goodness-of-fit (“are observed data consistent with a proposed biophysical model?”) or to interpret empirical biomarkers (“when observed dynamics change, what parameters are changing in the underlying system?”). Fitting parameters in biophysical models based on real data is typically an ill-posed problem because of the complexity of the models: many parameter settings would be consistent with a given noisy empirical timeseries. Nevertheless, some kinds of large-scale parameters may have systematic, tractable relationships to observed data. Here we propose a statistical framework based on Filtered Point Processes (FPP) to act as an interface between biophysical models and empirical neural power spectra. We show how the simple statistical framework can be used to facilitate interpretation of rhythmic, broadband, and cross-frequency coupling effects in empirical power spectra, and we show examples of dynamic parameters that can be fit using the framework. Ultimately, we intend for the framework to enable theoreticians to relate their models more closely to empirical data, capturing a broader range of statistical properties than prior methods.
14:50 - 15:25
Modelling Neural Activity During Adaptive Deep Brain Stimulation for Parkinson’s Disease
Madeleine M. Lowery
School of Electrical and Electronic Engineering, University College Dublin
Abstract
Deep brain stimulation (DBS) is an established surgical technique for treating the symptoms of Parkinson’s disease involving high frequency stimulation of neurons within the basal ganglia region of the brain. While very effective at treating motor symptoms including bradykinesia, akinesia and tremor, identification of the optimal stimulation settings for each patient is challenging. DBS can also result in stimulation-induced side effects due to stimulation of regions outside the target of interest. To address these limitations, new adaptive DBS methods are currently being investigated to automatically adjust stimulation parameters in real-time in response to changes in patient symptoms or side-effects. Adaptive or closed-loop stimulation requires the identification of biomarkers that are correlated with symptoms which can be modulated using appropriate control algorithms. Subthalamic beta band activity is correlated with the symptoms of bradykinesia and rigidity, and suppressed with both medication and DBS. It has emerged as a potential biomarker for closed-loop DBS, and along with other electrophysiological signatures, is currently being trialed as a biomarker for adaptive DBS in patients. Identification of the most appropriate control algorithms for adaptive DBS is similarly challenging, particularly as the exact therapeutic mechanisms of DBS and the neural pathways involved are not yet fully understood. Here we explore how biophysically accurate computational models can be used to develop and test new approaches for adaptive deep brain stimulation, including advanced closed-loop control algorithms. Models of the electric field surrounding the DBS electrode are coupled to individual neurons within the corticobasal ganglia network. Using the models, the effect of DBS on individual neural activity and extracellular potentials including local field potentials and electrocortiogram signals can be examined. In this way, computational models can help us understand the effects of electrical stimulation on neural activity at the single cell and system level, linking it to electrophysiological signals which capture population level activity and providing mechanistic insights into the system and its dynamics. We will show how different closed-loop DBS control algorithms can be implemented and tested in silico as part of the path to clinical translation.
16:00 - 16:35
Principles and Operation of Virtual Brain Twins
Meysam Hashemi
Aix Marseille Université, INSERM, INS, Inst Neurosci Syst, Marseille, France
Abstract
Current clinical methods often neglect individual variability by relying on population-wide trials, while mechanism-based approaches remain underutilized in neuroscience due to the brain’s complexity. This situation may change through the use of a Virtual Brain Twin (VBT), which is a personalized digital replica of an individual's brain, integrating structural and functional brain data into advanced computational models and inference algorithms. By bridging the gap between molecular mechanisms, whole-brain dynamics, and imaging data, VBT provides interpretable predictive capabilities, supporting clinicians in personalizing treatments and optimizing interventions. This talk presents the foundational components of VBT development, from anatomical modeling and simulation to Bayesian inference. Applications are illustrated across use cases including resting-state dynamics, healthy aging, multiple sclerosis, epilepsy, Parkinson’s disease, and Alzheimer’s disease.
16:35 - 17:10
The Human Neocortical Neurosolver as an interactive modeling tool to study the multi-scale mechanisms of human EEG/MEG signals
Katharina Duecker
Department of Neuroscience, Carney Institute for Brain Science, Brown University, RI, USA
Abstract
The Human Neocortical Neurosolver (HNN) is a user-friendly software tool developed to make biophysical modeling of the multiscale neural origin of Electroencephalography (EEG) and Magnetoencephalography (MEG) signals accessible to a broad audience. HNN is designed with a graphical user interface and workflows for the development and testing of hypotheses on the neural dynamics underlying E-/MEG measured brain oscillations and event-related potentials (ERPs). Similar functionality and tutorials can also be accessed through a Python-based API. The foundation of HNN is a laminated neocortical column model, consisting of excitatory pyramidal neurons and inhibitory interneurons in supragranular and infragranular layers. The model is activated by layer specific synaptic input from exogenous sources, including thalamus and higher order cortex. The primary current sources underlying E/MEG signals are calculated from the pyramidal neuron dendrites enabling one to one comparison to source localized signals. HNN’s model is explicitly designed with an intermediate level of biophysical detail that allows relatively computationally inexpensive simulations that can be run locally on a CPU. In this talk, I will provide an overview of the software and guidance on getting started with simulations of ERPs and brain rhythms. Furthermore, I will present a model update that reproduces pyramidal neuron dendritic calcium dynamics based on the biophysics of human cells. I will demonstrate that intracellular calcium is associated with large currents towards the cell body that have a dominating effect on the amplitude of E/MEG recordings.
The Organizers
Nicolò Meneghetti
Sant'Anna School of Advanced Studies, Pisa, Italy
PostDoc at the Computational Neuroengineering Lab
Publications: Google Scholar
Website: Link
Eleonora Russo
Sant'Anna School of Advanced Studies, Pisa, Italy
PI of the Brain Dynamics Laboratory
Publications: Google Scholar
Website: Link