NEURD offers automated proofreading and feature extraction for connectomics
Problem
Reconstructing neurons in millimeter-scale electron microscopy volumes requires laborious post-hoc proofreading to fix merge/split errors, and extracting morphological features demands stitching custom tool workflows.
Why it matters
Large error-free connectomes are needed to understand brain wiring, but manual proofreading is a major bottleneck.
What the AI did
NEURD, building on ML-based automated segmentation, decomposes meshed neurons into compact, extensively-annotated graph representations, offering automated proofreading and feature extraction.
Results
not stated in source
Impact
Evidence supports release of a software package for connectome proofreading; adoption not stated.
Record
| Venue | Nature |
|---|---|
| Published | 2025 |
| Work type | article |
| Authors | Brendan Celii, Stelios Papadopoulos, Zhuokun Ding, Paul G. Fahey, Eric Y. Wang, Christos Papadopoulos, Alexander B. Kunin, Saumil Surendra Patel et al. |
| Citations | 40 (OpenAlex, at import time) |
| References | 106 |
| Model or method | not stated in source |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1038/s41586-025-08660-5 |
| Open access | hybrid |
| Keywords | connectomics, electron microscopy, proofreading, segmentation, neuron morphology |
Importance rating
significant — thousands to millions affected
Proofreading is a major bottleneck for large error-free connectomes needed to understand brain wiring; tool aids a foundational neuroscience effort, but no scale stated.