Automated, Hue-Based Reconstruction of Large-Scale Neuronal Networks
Leiwe MN, Fujimoto S, Baba T, Moriyasu D, Saha B, Sakaguchi R, Inagaki S, Imai T. Automated neuronal reconstruction with super-multicolour Tetbow labelling and threshold-based clustering of colour hues. Nat Commun 2024;15(1):5279. doi: 10.1038/s41467-024-49455-y.
Background: Mapping densely labelled neuronal circuits is limited by the difficulty of distinguishing overlapping neurites using light microscopy and by the laborious nature of manual tracing. Conventional multicolour fluorescence methods such as Brainbow and Tetbow use only three fluorescent proteins, producing too few colour variations to separate neighbouring neurons effectively.
Hypothesis: This study hypothesized that combining super-multicolour Tetbow labelling using more than three fluorescent proteins with an automated hue-based analysis pipeline could accurately reconstruct densely labelled neuronal circuits without relying on physical continuity.
Methods: The authors developed the QDyeFinder pipeline, which uses spectral unmixing and quantitative analysis of seven fluorescent proteins to extract neurite colour vectors. Neurolucida 360 was employed for soma detection and automated neurite tracing, generating fragments that were clustered by colour similarity using the custom dCrawler algorithm. Validation involved manual reconstructions for ground truth comparisons in cortical and olfactory bulb samples.
Results: Seven-colour Tetbow labelling provided superior neuronal discriminability (>99.9%) compared to conventional three-colour methods. Automated detection identified over 15,000 neurite fragments that were accurately clustered into individual neurons at an optimal threshold (Th(d) = 0.2). QDyeFinder successfully reconstructed dendritic and axonal morphologies, including neurites spanning multiple brain sections.
Conclusions: Super-multicolour labelling combined with the QDyeFinder pipeline enables fully automated, hue-based neuronal reconstruction across large volumes, overcoming limitations of continuity-dependent tracing and significantly advancing scalable connectomics.
