Deep learning on graphs for morphofunctional analysis and comparison of brains

Summary

The development of brain imaging methods has generated a considerable amount of morphological and functional data. However, their exploration and comparison over time for an individual (development and aging), between individuals (variability within the species), and even more so between different species have been only partial. We propose to model these data in the form of graphs, then to use recent approaches of artificial intelligence to better analyze them.

This approach has already been initiated by a multidisciplinary consortium of researchers in neuroanatomy, animal biology and computer science as well as neurosurgeons during the NeuroGeo and Neuro2Co regional projects (LIFAT, INRAE, INSERM). It led to the creation of SILA3D, an open access software platform allowing the representation of anatomo-functional data in the form of graphs based on the interactive semantic segmentation of images.

In this context, the proposed thesis aims to create new algorithms for anatomo-functional analysis and comparison of brain data using classic methods (graph theory) but also more recent ones (deep neural networks on graphs (GNN), geometric deep learning ...).
The general objectives of this thesis are:

• To specify different strategies for modeling data as graphs. For this, morphological and functional data from different imaging modalities, including structural MRI and tractography, will be combined using different approaches to be defined. The PhD student will use two data sets already acquired: in vivo MRI of growing lambs (PRC and PIXANIM).
• To study the differences between individuals and over time (follow-up to the brain development of the lamb from birth to adulthood). The PhD student will propose several graph comparison methods exploiting recent advances in Deep Learning on Graphs (GNN).

The scientific challenges associated with these objectives are (1) to develop new graph-based deep learning methods for the detection and classification of particular substructures in an encephalon (semi-supervised classification of nodes); (2) to develop new graph-based deep learning methods for the comparison, discrimination, and classification of encephalon (supervised or unsupervised classification).

Publications

• Bourlier, A., Chaillou, E., & Ramel, J. Y. (2023, September). 3DBrainMiner, a tool to generate brain graphs from MR-images. In NeuroCompare: Comparative Neuronal Circuits For Adaptive Behaviour.
• Bourlier, A., Chaillou, E., Ramel, J. Y., & Slimane, M. Comparison Between CNN and GNN Pipelines for Analysing the Brain in Development. Proceedings Copyright, 475, 482.
• Achin, I., Morisse, M., Parias, C., Love, S. A., Bourlier, A., Lasserre, O., ... & Chaillou, E. (2023, May). Are sheep really afraid of the wolf?. In 16. International Meeting of the French Neuroscience Society.