Generating synthetic task-based brain fingerprints for population neuroscience using deep learning

作者：Emin Serin, Kerstin Ritter, Günter Schumann, Tobias Banaschewski, André F. Marquand, Henrik Walter, Gunter Schumann, Andreas Heinz, Markus Ralser, Sven Twardziok, Nilakshi Vaidya, Antoine Bernas, Emin Serin, Marcel Jentsch, Esther Hitchen, Elli Polemiti, Hedi Kebir, Tristram A. Lett, Jean-Charles Roy, Roland Eils, Ulrike H. Taron, Tatjana Schütz, Kerstin Schepanski, Karina Jansone, Nina Christmann, Andreas Meyer‐Lindenberg, Heike Tost, Nathalie Holz, Emanuel Schwarz, Argyris Stringaris, Maja Neidhart, Frauke Nees, Beke Seefried, Rieke Aden, Ole A. Andreassen, Lars T. Westlye, Dennis van der Meer, Sara Fernández, Rikka Kjelkenes, Helga Ask, Michael A. Rapp, Mira Tschorn, Sarah Jane Böttger, Gaia Novarino, Mel Slater, Guillem Feixas, Francisco José Eiroá‐Orosa, Reiya Itatani, Jaime Gallego, Álvaro Pastor, Andreas J. Forstner, Per Hoffmann, Markus M. Nöthen, Isabelle Claus, Abigail J. Miller, Carina M. Mathey, Stefanie Heilmann‐Heimbach, Peter Sommer, Myrto Patraskaki, Johannes Wilbertz, Karen Schmitt, Viktor Jirsa, Spase Petkoski, Anastasios-Polykarpos Athanasiadis, Charlie Pearmund, Bernhard Spanlang, Sören Hese, Paul Renner, Tianye Jia, Yunman Xia, Jiacan Yuan, Yuxiang Dai, Yuzhu Li, Yanqing Zhang, Xiao Chang, Vince D. Calhoun, Ameli Schwalber, Venessa Köhler, Paul Thompson, Nicholas Clinton, Sylvane Desrivières, Di Chen, Kofoworola Agunbiade, Zuo Zhang, Xinyang Yu, Allan H. Young, Tamara Schikowski, Ragnhild Eek Brandlistuen, Bernd Carsten Stahl, George Ogoh · 发表于：Communications Biology · 年份：2025 · DOI：10.1038/s42003-025-09158-6 · 被引用次数：1 · 研究领域：Functional Brain Connectivity Studies、EEG and Brain-Computer Interfaces、Advanced Memory and Neural Computing

Task-based functional magnetic resonance imaging (fMRI) reveals individual differences in neural correlates of cognition but faces scalability challenges due to cognitive demands, protocol variability, and limited task coverage in large datasets. Here, we propose DeepTaskGen, a deep-learning approach that synthesizes non-acquired task-based contrast maps from resting-state (rs-) fMRI. We validate this approach using the Human Connectome Project lifespan data, then generate 47 contrast maps from 7 different cognitive tasks for over 20,000 individuals from UK Biobank. DeepTaskGen outperforms several benchmarks in generating synthetic task-contrast maps, achieving superior reconstruction performance while retaining inter-individual variation essential for biomarker development. We further show comparable or superior predictive performance of synthetic maps relative to actual maps and rs-connectomes across diverse demographic, cognitive, and clinical variables. This approach facilitates the study of individual differences and the generation of task-related biomarkers by enabling the generation of arbitrary functional cognitive tasks from readily available rs-fMRI data.