Automated Soma Detection in Whole-Brain Imaging via Post-Tracing Multi-Furcation Morphometry

Accurate single-neuron morphology reconstruction is essential for understanding brain structure and function. A critical step in this process is the detection of somas in 3D brain images, which remains a challenge due to the complex structure of the brain and the large amount of whole brain imaging...

Full description

Saved in:
Bibliographic Details
Main Author: Gu Xiaoqin
Format: Article
Language:English
Published: EDP Sciences 2025-01-01
Series:BIO Web of Conferences
Online Access:https://www.bio-conferences.org/articles/bioconf/pdf/2025/25/bioconf_icbb2025_03022.pdf
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Accurate single-neuron morphology reconstruction is essential for understanding brain structure and function. A critical step in this process is the detection of somas in 3D brain images, which remains a challenge due to the complex structure of the brain and the large amount of whole brain imaging data [1]. Despite decades of research, automated soma detection methods continue to struggle with difficulties arising from non-spherical soma morphology, imaging artifacts, and variability in fluorescence labelling [2]. Faced with these constraints, traditional pipelines rely on manual annotation or spherical assumptions, which are error-prone and operator dependent. In this study, we propose a novel, automated pipeline that leverages local morphometry features, particularly multi-furcation clusters, to detect soma after initial fiber tracing in high- resolution, large-scale whole mouse brain images. Our method eliminates the need for spherical priors, accounts for anisotropy in the z-axis resolution, and operates without human intervention. Validated on 253 public annotated mouse brain datasets, the pipeline achieved 99.2% accuracy in localizing soma within 128³ voxel blocks centered on ground-truth positions. This pipeline provides a robust, high-throughput solution for whole-brain neuronal reconstruction and represents a step forward in automated neuroscientific analysis.
ISSN:2117-4458