In silico voltage-sensitive dye imaging reveals the emergent dynamics of cortical populations

Taylor Newton, Michael Reimann, Marwan Abdellah, Grigori Chevtchenko, Eilif Müller, Henry MarkramNature Communications 2021 · DOI

Abstract

Voltage-sensitive dye imaging (VSDI) is a powerful technique for interrogating membrane potential dynamics in assemblies of cortical neurons, but with effective resolution limits that confound interpretation. To address this limitation, we developed an in silico model of VSDI in a biologically faithful digital reconstruction of rodent neocortical microcircuitry. Using this model, we extend previous experimental observations regarding the cellular origins of VSDI, finding that the signal is driven primarily by neurons in layers 2/3 and 5, and that VSDI measurements do not capture individual spikes. Furthermore, we test the capacity of VSD image sequences to discriminate between afferent thalamic inputs at various spatial locations to estimate a lower bound on the functional resolution of VSDI. Our approach underscores the power of a bottom-up computational approach for relating scales of cortical processing.

Introduction

Voltage-sensitive dye imaging (VSDI) is a cornerstone of modern neuroscience, allowing researchers to observe waves of electrical activity across the brain's surface in real-time. However, the technique suffers from inherent "blurriness" due to light scattering and uneven dye penetration. A single pixel represents an average of thousands of neural components across multiple layers, leading to long-standing debates over which neurons drive the signal and whether it captures individual spikes or slower subthreshold activity. Without a ground truth for comparison, interpreting VSDI involves significant uncertainty.

Methods: The Digital Twin Approach

To resolve these ambiguities, we developed an in silico model of VSDI using the Blue Brain Project’s digital reconstruction of the rodent neocortical microcircuit (NMC). This model includes approximately 31,000 morphologically detailed neurons and 37 million synapses, representing a biologically faithful reconstruction based on decades of anatomical data.

The simulation workflow, implemented via the EMSim software, integrated three layers: electrophysiology (modeling the firing of 31,000 neurons), optical physics (using Monte Carlo simulations to model photon travel and scattering), and microscope optics (applying a ray-transfer model to simulate a real-world tandem-lens microscope).

Schematic of the Monte Carlo simulation for modeling photon scattering and absorption within the digital cortical column.

This "digital twin" allows for the generation of synthetic VSDI movies where the underlying neural activity is perfectly known, providing the missing ground truth for interpretation.

Results

Cellular Origins and the "Effective Surface Area" Rule

The model reveals that the VSDI signal is primarily driven by neurons in Layers 2/3 and 5. This dominance is attributed to their large dendritic trees, which provide a significant surface area for dye binding—an "effective surface area" rule. Over 90% of raw fluorescence originates from the top 500μm of the cortex, with Layer 5 neurons contributing up to 67% of the signal during evoked activity, despite their deeper location.

Temporal Dynamics and Spike Transparency

A critical finding is that VSDI does not capture individual action potentials. Instead, the signal is almost entirely driven by subthreshold membrane potential changes—the "inputs" rather than "outputs" of neurons. Furthermore, peak population spiking occurs approximately 7 milliseconds before the VSDI signal peaks, indicating that VSDI acts as a delayed "echo" of synaptic activity.

Spatiotemporal evolution of the simulated VSDI signal (ΔF/F₀) following a stimulus, showing the expanding wave of cortical activity.

Functional Resolution Limits

By using machine learning (XGBoost) to distinguish between different input locations, the study identified a functional resolution limit of approximately 25 micrometers. Below this threshold, optical blur renders distinct inputs indistinguishable, establishing a lower bound for VSDI precision.

Discussion

This bottom-up computational approach demonstrates that digital reconstructions can effectively "de-noise" complex biological signals. By bridging the gap between microscopic activity and mesoscale imaging, this model provides a framework for more accurate interpretation of experimental VSDI data and highlights the importance of considering cellular morphology and laminar structure in brain imaging.

Summary updated

Abstract

Voltage-sensitive dye imaging (VSDI) is a powerful technique for interrogating membrane potential dynamics in assemblies of cortical neurons, but with effective resolution limits that confound interpretation. To address this limitation, we developed an in silico model of VSDI in a biologically faithful digital reconstruction of rodent neocortical microcircuitry. Using this model, we extend previous experimental observations regarding the cellular origins of VSDI, finding that the signal is driven primarily by neurons in layers 2/3 and 5, and that VSDI measurements do not capture individual spikes. Furthermore, we test the capacity of VSD image sequences to discriminate between afferent thalamic inputs at various spatial locations to estimate a lower bound on the functional resolution of VSDI. Our approach underscores the power of a bottom-up computational approach for relating scales of cortical processing.
BibTeX
@inproceedings{newton2021silico,
  author = {Taylor Newton and Michael Reimann and Marwan Abdellah and Grigori Chevtchenko and Eilif Müller and Henry Markram},
  title = {{In silico voltage-sensitive dye imaging reveals the emergent dynamics of cortical populations}},
  booktitle = {Nature Communications},
  year = {2021},
  doi = {10.1038/s41467-021-23901-7},
}
Figure