NEURD offers automated proofreading and feature extraction for connectomics

Area of Science:

• Neuroscience
• Computational Biology
• Bioinformatics

Background:

• Millimeter-scale electron microscopy (EM) volumes at nanometer resolution are now common.
• Machine learning (ML) advances enable dense reconstruction of cellular compartments.
• Automated segmentation requires post-hoc proofreading for accurate connectomes.

Purpose of the Study:

• To present NEURD, a software package for analyzing neural morphology and connectivity.
• To enable efficient extraction of detailed morphological information from 3D neuronal meshes.
• To automate downstream analyses of large-scale EM datasets.

Main Methods:

• Developed NEURD software based on open-source mesh manipulation tools.
• Decomposed meshed neurons into compact, annotated graph representations.
• Automated proofreading, cell classification, spine detection, and proximity analyses using feature-rich graphs.

Main Results:

• NEURD successfully converts complex 3D neuronal meshes into usable graph structures.
• Automated tasks include state-of-the-art merge error correction and detailed morphological feature extraction.
• The software facilitates analysis of neural morphology and connectivity in large EM datasets.

Conclusions:

• NEURD enhances accessibility to massive neural datasets for neuroscience researchers.
• Automated graph-based analysis streamlines complex morphological and connectivity studies.
• The software supports diverse research questions in neural morphology and connectomics.