Inferring Pyramidal Neuron Morphology using EAP Data.

Area of Science:

• Computational neuroscience
• Neuroimaging

Background:

• Understanding neuron position and morphology is crucial for interpreting brain activity.
• Current methods for inferring neuronal structure from electrophysiological data are limited.

Purpose of the Study:

• To develop and validate a computational algorithm for inferring cortical pyramidal neuron position and morphology from extracellular action potential recordings.
• To create a generic, stylized pyramidal neuron model adaptable to different cortical layers and rodent motor cortex neuron types.

Main Methods:

• Developed a generic pyramidal neuron model with adjustable parameters for soma location, dendrite morphology, and orientation.
• Employed a machine learning approach, specifically a convolutional neural network, trained on simulated local field potentials.
• Utilized spatio-temporal extracellular action potential (EAP) waveform profiles for parameter prediction.

Main Results:

• The algorithm demonstrated reliable inference of key neuron position and morphology parameters from simulated EAP waveforms.
• Preliminary validation using in vivo data provided partial support for the algorithm's efficacy.
• The study successfully mimicked realistic electrophysiological dynamics of pyramidal cells.

Conclusions:

• The proposed inverse modeling scheme and machine learning approach offer a promising method for inferring neuronal structure from electrophysiological recordings.
• Ongoing work focuses on automating the inference pipeline for broader application in neuroscience research.
• This computational tool has the potential to advance our understanding of neural circuits and function.