Linking structure and function is a fundamental principle when aiming to describe how information is processed in the brain, and the most accurate results are obtained by studying these in situ. While the subcellular structures underlying computations in cortical neurons are below 1 um in diameter, an entire cell can extend processes across more than 1,000,000 um3 of tissue. Here we present a workflow to analyse large neuronal morphologies at high resolution in 3D and some of the challenges during acquisition and image processing due to their size and complexity. Neurons were passively filled with biocytin during electrophysiological recording, then fixed and co-localized with a fluorescently tagged streptavidin to label their processes. Spinning disk confocal microscopy (3i Yokogawa CSU-W1) was used to acquire the volume of tissue filled by these cells at a high resolution, often generating datasets with over 10,000 individual images and larger than 100GB per neuron. Individual fields of view were deconvolved using Huygens Professional (Scientific Volume Imaging) to remove out-of-focus signal and improve resolution in the raw data, and stitched to align spatially and remove overlapping tile regions. Neurons were then traced in 3D across these image datasets via the user-guided method in Neurolucida® 360 Studio software (MBF Bioscience). Passing such large datasets through these computationally intense tasks demanded dedicated resources, specific file format conversions and several data transfers. Despite the many challenges raised during this project, this framework will enable the discovery of mechanisms linking neuronal structure and function in the human brain.
University of Queensland