Exploring Consistency of Connectome Across Species Using Inter-Bundle Topological Graph Network (ibTGN)

Identifying anatomically consistent structures across species is crucial for understanding brain evolution, cognition, and behavior. However, registration-based methods become time-consuming and inaccurate when significant differences in size and morphology exist between the brains of different species. In this work, we developed a novel method that aimed to characterize inter-bundle topology and enable the identification of shared patterns of white matter connectivity between human and non-human primate brains. We encoded the whole-brain tractography as a graph with streamlined local and global features. Fiber similarity across species was calculated based on low-dimensional features in latent space and trained on cross-species datasets. Locally and morphologically equivalent fiber bundles were then identified across species. Our experimental results demonstrate that this proposed method enables consistent identification of anatomical pathways without the need for complex cross-species-based registration, thereby providing valuable insights into the evolutionary conservation of neural circuitry underlying cognitive processes. The project code is available at https://github.com/secuy/ibTGN.git.