High-Resolution Single-Neuron Reconstruction Analysis in Golgi-Stained Brain Tissues

Understanding the structural and functional organisation of brain networks is a fundamental objective in neuroscience, with three-dimensional (3D) reconstruction of single-neuron morphology serving as a critical foundation. The Golgi staining method, which enables random neuronal labeling and provides high-contrast signals in both optical and X-ray microscopy, remains a valuable tool for morphological analysis. However, its widespread application in large-scale neuronal reconstructions is hindered by signal discontinuities in neuronal branches, high-density labeling, and complex background interference. While automated reconstruction methods perform well in sparsely labelled and morphologically simple neuronal populations, their effectiveness is limited in Golgi-stained samples. Here we develop a semi-automated single-neuron reconstruction method for Golgi-stained mouse brain neurons (SNR-Golgi). By integrating three key technical modules-background denoising, single-neuron extraction, and branch repair-SNR-Golgi significantly enhances the accuracy and completeness of neuronal reconstruction. In fluorescence micro-optical sectioning tomography (fMOST) datasets, SNR-Golgi demonstrated superior performance in neuronal reconstruction within the mouse somatosensory cortex, achieving a 30% increase in reconstructed branch count, a 76% improvement in total branch length, and a 3.7-fold increase in axonal length. Additionally, in synchrotron-based X-ray imaging datasets, SNR-Golgi enabled submicron-resolution 3D reconstruction of single neurons. These results demonstrate that SNR-Golgi effectively addresses the complexity of Golgi-stained samples and provides robust technical support for the structural analysis of brain neurons across various imaging modalities.