Localising and classifying neurons from high density MEA recordings.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Electrophysiology

Background:

• Neuronal microcircuits comprise diverse, spatially organized cell classes crucial for function.
• Spatial information in neural circuit characterization is underutilized, especially from multi-electrode recordings.
• Current methods primarily exploit temporal data (spike shape, firing patterns) for neuron classification.

Purpose of the Study:

• To develop and validate methods for localizing and classifying neurons using spatial patterns from extracellular signals.
• To demonstrate the utility of spatial signal patterns for inferring neuronal morphology and improving circuit analysis.
• To address limitations of existing methods in multi-electrode recording scenarios (distances < 60μm).

Main Methods:

• Utilized current source models to link extracellular potential generation to neuronal electrophysiology and morphology.
• Applied spatial patterns across multi-electrode arrays to develop neuron localization and classification algorithms.
• Validated models using simulated data and subsequently applied them to experimentally recorded data.

Main Results:

• Achieved low fitting errors in localizing and classifying neurons using simulated data.
• Demonstrated correspondence between localization statistics and expected recording radii in experimental data.
• Provided evidence supporting the separation of neurons into putative morphological classes based on spatial patterns.

Conclusions:

• Extracellular recordings contain exploitable spatial information for neuron localization and classification.
• The developed spatial methods complement existing temporal-based approaches for comprehensive neuron characterization.
• This approach enhances the understanding of neuronal microcircuits by integrating spatial and functional criteria.