1. Background
Magnetic Resonance Imaging (MRI) provides an invaluable way to measure brain connectivity, which can be defined as either structural: linking regions that are connected by physical tracts or structurally similar, or functional: linking regions with similar activity over time. Functional brain connectivity has been suggested to be more predictive of later psychotic illness than brain structure (Morgan et al., BP:CNNI 2021), however structural brain images are easier to obtain from patient populations.
2. Novelty & Importance
Here we take a multimodal approach to explore whether structural brain connectivity can be used to predict functional connectivity, and then ascertain whether the resulting generated functional connectivity can help predict outcomes for patients with psychotic illnesses. To that end, we will use our new Morphometric Inverse Divergence (MIND) method to estimate structural connectivity, which showed greater sensitivity to inter-individual differences than previous approaches (Sebenius et al., Nature Neuroscience 2023). We will also build on a growing literature of tools to generate functional connectivity from structural brain images (Jamison et al., Nature Methods 2025; Zalesky et al., Network Neuroscience 2024; Mhiri et al., IPMI 2021). Finally, we will work closely with clinicians and people with lived experience of psychosis to identify how these tools might be used in practice.
3. Aims and Objectives
Our aims are: 1) to build a generative AI pipeline to predict functional connectivity from MIND structural brain connectivity. 2) To explore whether the generated functional connectivity or the difference between generated and actual functional connectivity can predict outcomes for patients with psychotic illnesses. 3) To package the tool to generate functional connectivity from MIND, for use by other, multidisciplinary researchers. Ultimately, being able to predict likely outcomes and treatment response ahead of time could help target treatments to patients with poor prognosis.
References:
Sebenius et al., Nature Neuroscience, 2023 https://doi.org/10.1038/s41593-023-01376-7
Morgan et al., Biological Psychiatry CNNI, 2021 https://doi.org/10.1016/j.bpsc.2020.05.013
Jamison et al., Krakencoder: a unified brain connectome translation and fusion tool, Nature Methods, 2025, https://doi.org/10.1038/s41592-025-02706-2
Zalesky et al., Network Neuroscience 2024, https://doi.org/10.1162/netn_a_00400
Mhiri et al., IPMI 2021, https://doi.org/10.1007/978-3-030-78191-0_16