Petkantchin, RemyBerchet, AdrienPeng, HanchuanMarkram, HenryKanari, Lida2025-01-072025-01-072025-01-062024-10-0410.1101/2024.10.04.616605https://infoscience.epfl.ch/handle/20.500.14299/242556Recent experimental advancements, including electron microscopy reconstructions, have produced detailed connectivity data for local brain regions. On the other hand, for inter-regional connectivity, large-scale imaging techniques such as MRI are best suited to provide insights. However, understanding the relationship between local and long-range connectivity is essential for studying both healthy and pathological conditions of the brain. Leveraging a novel dataset of whole-brain axonal reconstructions, we present a technique to predict whole-brain connectivity at single cell level by generating detailed whole-brain axonal morphologies from sparse experimental data. The computationally generated axons accurately reproduce the local and global morphological properties of experimental reconstructions. Furthermore, the computationally synthesized axons generate large-scale inter-regional connectivity, defining the projectome and the connectome of the brain, thereby enabling the in silico experimentation of large brain regions.encomputational synthesisaxonal projectionsbrain connectivitydata-driven clusteringartificial neural networktext::preprint