Automated neuronal reconstruction with super-multicolour Tetbow labelling and threshold-based clustering of colour hues

Abstract

Abstract Fluorescence imaging is widely used for the mesoscopic mapping of neuronal connectivity. However, neurite reconstruction is challenging, especially when neurons are densely labelled. Here, we report a strategy for the fully automated reconstruction of densely labelled neuronal circuits. Firstly, we establish stochastic super-multicolour labelling with up to seven different fluorescent proteins using the Tetbow method. With this method, each neuron is labelled with a unique combination of fluorescent proteins, which are then imaged and separated by linear unmixing. We also establish an automated neurite reconstruction pipeline based on the quantitative analysis of multiple dyes (QDyeFinder), which identifies neurite fragments with similar colour combinations. To classify colour combinations, we develop unsupervised clustering algorithm, dCrawler, in which data points in multi-dimensional space are clustered based on a given threshold distance. Our strategy allows the reconstruction of neurites for up to hundreds of neurons at the millimetre scale without using their physical continuity.