Back

Brain Diffusion Transformer for Personalized Neuroscience and Psychiatry

Zhai, R.; Hu, Y.; Zheng, L.; Xiang, S.; Xie, C.; Peng, L.; Banaschewski, T.; Barker, G. J.; Bokde, A. L. W.; Bruehl, R.; Desrivieres, S.; Flor, H.; Garavan, H.; Gowland, P.; Grigis, A.; Heinz, A.; Lemaitre, H.; Martinot, J.-L.; Martinot, M.-L. P.; Artiges, E.; Nees, F.; Orfanos, D. P.; Poustka, L.; Smolka, M. N.; Hohmann, S.; Holz, N.; Vaidya, N.; Whelan, R.; Zhang, Z.; Robinson, L.; Winterer, J.; King, S.; Zhang, Y.; Kebir, H.; Schmidt, U.; Sinclair, J.; Stringaris, A.; Schumann, G.; Bzdok, D.; Walter, H.; Rolls, E. T.; Sahakian, B.; Robbins, T. W.; Feng, J.; Gong, W.; Jia, T.; IMAGEN Consort

2025-04-14 neuroscience
10.1101/2025.04.12.648506 bioRxiv
Show abstract

Task-fMRI analyses typically focus on localized activation contrasts between stimuli, neglecting the brains dynamic hierarchy. We introduce Brain Diffusion Transformer (Brain-DiT), a deep generative model capturing recurrent processing underlying individualized neurocognitive state transitions via functional networks. Without prior assumptions, Brain-DiT identifies canonical cognitive regions in the brain and reveals replicable subgroups with distinct neural circuits in large cohorts, offering critical clinical insights overlooked by traditional methods: individuals exhibiting negative emotion bias, linked to language-related regions, had a 12-fold higher likelihood of major depression, and those with maladaptive inhibition strategies, associated with overactive medial frontal regions, showed a 9-fold increased risk of alcohol abuse. By bridging cognitive theory and psychiatric applications, Brain-DiT provides a unified analytical paradigm, paving the way for operational personalized medicine in psychiatry.

Matching journals

The top 4 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.