Year: 2026
Type: article
Abstract: Quantitative analysis of Golgi-stained neurons is fundamentally limited by non-specific precipitates that introduce substantial bias into morphological measurements. Conventional denoising approaches, designed mainly for low-resolution medical imaging or reconstruction artefacts, are insufficient for suppressing this chemically induced noise at sub-micrometre resolution. Here, we present a high-fidelity denoising strategy for Golgi-stained neurons (HiFi Golgi), a bias-controlled analytical strategy for three-dimensional (3D) Golgi-stained neuronal imaging. We established a multidimensional evaluation framework to systematically quantify the trade-off between noise suppression and structural fidelity, enabling rigorous comparison of candidate filtering strategies. Using this framework, an optimised low-parameter bilateral filter was identified that minimises noise-induced bias while preserving neuronal morphology. The HiFi Golgi strategy consistently reduces bias across diverse brain regions, imaging resolutions, and model species without parameter re-optimization, improving the reliability and reproducibility of single-neuron reconstruction. This work provides a generalizable and high-fidelity analytical approach for bias-controlled volumetric imaging, offering a reproducible framework for quantitative morphological analysis in complex neuronal systems.
Source: Cell Proliferation
DOI: 10.1111/cpr.70291
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $3,680
OpenAlex ID: W7229020876
