NeuroRefiner: Morphology-Aware Multi-Agent Refinement for 3D Fluorescence Microscopy Neuron Segmentation

2026-08-10

AI Summary

Existing methods for 3D fluorescence microscopy image segmentation often fail to simultaneously preserve neuronal local details and global topology, resulting in fragmented reconstructions. To address this challenge, this work proposes NeuroRefiner—the first framework to incorporate a multi-agent collaborative mechanism into neuron segmentation—mimicking an expert’s iterative editing process. Three cooperative agents diagnose topological errors, generate correction instructions, and validate reconstruction quality, respectively, while a dedicated network, TopoRefineNet, executes multi-round voxel-level refinements. By integrating morphology-aware processing, cross-modal features, and interpretable iterative corrections, NeuroRefiner consistently outperforms state-of-the-art methods across the BigNeuron, CWMBS, and ZBFWB benchmarks, achieving a 3.02% F1-score improvement on ZBFWB and significantly enhancing topological accuracy.

Abstract

Accurate 3D neuron segmentation in fluorescence microscopy is critical for neuroscience. However, the sparse and elongated morphology of neurons poses significant challenges to existing segmentation methods. These methods struggle to preserve both local details and global topology, leading to fragmented results. To address this, we propose NeuroRefiner, a multi-agent system that formalizes the human expert workflow involving iterative global observation and local editing. Specifically, NeuroRefiner comprises three collaborative agents dedicated to diagnosing topological errors, generating correction instructions, and validating refinement quality. To facilitate agent instruction-guided segmentation refinement, we propose TopoRefineNet, a dedicated 3D U-Net-based tool that leverages cross-modality feature fusion to generate refined masks. Through multi-round agent reasoning and voxel-level editing, NeuroRefiner produces topologically more accurate segmentations with enhanced interpretability. Experiments on the BigNeuron, CWMBS, and ZBFWB datasets demonstrate that NeuroRefiner outperforms state-of-the-art methods, notably achieving a 3.02% improvement in F1 score on the challenging ZBFWB dataset.

Problem

• neuron segmentation
• 3D fluorescence microscopy
• morphology
• topological accuracy
• sparse structures

Innovation

• multi-agent refinement
• morphology-aware segmentation
• topological error correction
• 3D fluorescence microscopy
• TopoRefineNet