Probe-EM: Targeted Neuron Tracing via Training-Free Semantic Verification

2026-07-06

AI Summary

This work addresses the excessive manual proofreading burden caused by oversegmentation in electron microscopy-based neuron reconstruction by proposing a training-free, targeted tracing framework. The method integrates skeleton-guided heuristic spatial search with a probe-and-verify loop, introducing a novel training-agnostic semantic validation mechanism. It further combines planar ensemble consensus with axial spatiotemporal propagation and incorporates dimension-aware semantic verification alongside interactive visualization techniques. Evaluated in a zero-shot setting, the approach achieves highly robust segmentation correction, significantly outperforming supervised baselines and enabling efficient human-in-the-loop collaboration within Neuroglancer, thereby reducing manual proofreading time by 33.4%.

Abstract

Establishing large-scale, high-resolution neural connectivity maps is fundamental to elucidating the structural basis of brain function. However, when processing terabyte- or petabyte-scale electron microscopy data, over-segmentation inherent in automated reconstruction algorithms remains a critical bottleneck, requiring extensive manual proofreading spanning person-years. To alleviate the heavy reliance on annotated data and the limited flexibility of conventional tracing methods, we propose a training-free, targeted neuron tracing framework. Specifically, we introduce a skeleton-guided Heuristic Spatial Search paradigm that leverages geometric priors to iteratively reconstruct neuronal morphologies through a probing-verification cycle. To achieve robust zero-shot semantic verification, we further develop a Dimension-Aware Semantic Verification strategy built upon the foundation model NeuroSAM 2. This strategy resolves intra-slice splits via Planar Ensemble Consensus and inter-slice splits via Axial Spatio-Temporal Propagation. Notably, we integrate the proposed workflow into the Neuroglancer visualization platform, enabling an interactive human-in-the-loop proofreading system. Experimental results demonstrate that the proposed method outperforms supervised baselines and reduces manual proofreading time by 33.4%. The source code is publicly available at https://github.com/HeadLiuYun/Probe-EM.

Problem

• neuron tracing
• over-segmentation
• electron microscopy
• manual proofreading
• neural connectivity

Innovation

• training-free
• semantic verification
• neuron tracing
• geometric priors
• human-in-the-loop