Brain function cannot be fully characterized by any single modality. Instead, each imaging technique offers a distinct balance of temporal and spatial resolution (Zhang et al., 2020). EEG and MEG capture millisecond-level neural dynamics, whereas fMRI and PET offer greater spatial specificity and sensitivity to hemodynamic or molecular processes. This fundamental trade-off has long shaped how researchers investigate fast neural dynamics vs. large-scale network organization (Horwitz and Poeppel, 2002). Over the last two decades, multimodal neuroimaging has emerged as a powerful strategy to bridge these limitations. Simultaneous acquisitions—such as EEG/fMRI or MEG/fMRI—together with computational fusion frameworks, including Bayesian and other generative or model-based approaches, now enable more comprehensive and biologically-grounded characterizations of brain activity (He and Liu, 2008; Trujillo-Barreto et al., 2023).

From a structural point of view, additional integration is possible, from structure-function relationships to across-scale integration: Microstructural organization provides a bridge to mesoscale and macroscale organization and can constrain functional dynamics in integrative models (Amunts and Zilles, 2015; Honey et al., 2009).

Further, multimodal brain science increasingly extends beyond imaging to genetic, behavioral, and clinical variables, pushing the field from pairwise associations toward multivariate and machine-learning methods tailored to high-dimensional, heterogeneous datasets (Liu and Calhoun, 2014; Bogdan et al., 2017). Importantly, such incorporation of individual biomedical data from various domains informs clinical decision-making and supports the development of patient-specific interventions (Pham et al., 2024).

This Research Topic, Multimodal brain data integration and computational modeling, was conceived to: (i) Explore strategies for the effective fusion of heterogeneous neural data; (ii) Present computational and machine-learning approaches for modeling complex brain processes; and (iii) Demonstrate applications in basic and clinical neuroscience. The included articles are organized thematically below. Across the Research Topic, a recurring message is that integration is not only across modalities, but also across levels of description—from genes and microstructure to systems-level brain function.

A first set of contributions emphasizes multilevel integration, combining molecular, cellular, and microstructural measures with systems-level functional organization. Nazac et al. review recent advances in super-resolution microscopy and deep learning-based segmentation for neuronal microstructure, highlighting how modern computational pipelines enable quantitative characterization of features such as dendrites and spine morphology from high-resolution imaging modalities (e.g., SIM, STED, STORM, and MINFLUX). Piszczek et al. propose an integrative workflow that combines evolutionary genetic data with searchable, psychologically organized fMRI resources, using a genetic algorithm for biclustering (GABi) to identify gene sets under selection whose expression profiles relate to functional networks. Together, these studies underscore that progress in multimodal neuroscience depends not only on acquiring diverse data, but also on building analysis frameworks that connect biological mechanisms to interpretable, system-level models across scales.

A second set of contributions advances multimodal integration by learning cross-modal mappings that mitigate practical constraints such as data scarcity, cost, and radiation exposure. Revathy and Karthiga introduce AutismSynthGen, a privacy-preserving framework for autism spectrum disorder (ASD) prediction that synthesizes multimodal ASD data—including structural MRI, EEG, behavioral measures, and severity scores—using a transformer-based generative model with differential privacy coupled with an adaptive mixture-of-experts ensemble to improve predictive performance. Ferrante et al. propose a deep-learning approach to synthesize TSPO-PET images from structural MRI, aiming to approximate PET-derived neuroinflammation-related information from a more widely available modality; the generated images reproduce key spatial signal patterns and contrast properties observed in real TSPO-PET scans. Together, these studies illustrate how cross-modal learning and synthesis can make multimodal assessments more feasible and scalable, while preserving clinically relevant information when direct acquisition is limited.

A third theme focuses on robust inference and prediction, addressing noise, dimensionality, and heterogeneity in electrophysiology and connectivity. Patel et al. present a hybrid framework (DWT-CNN-BiGRU) for the classification of EEG signals from alcoholic and control subjects, combining wavelet-based denoising with convolutional and recurrent components to capture spatial and temporal structure and improve performance in a clinically relevant task. Ming and Kundu introduce a semi-parametric SVM approach that leverages high-dimensional brain network features, integrating static and dynamic functional connectivity across single and multiple fMRI experiments and both resting-state and task-based datasets; Their evaluation across independent datasets indicates improved classification relative to the tested baselines. Together, these contributions highlight how careful preprocessing and principled statistical learning can strengthen generalizable prediction from noisy, high-dimensional brain signals—an essential requirement for scalable multimodal modeling.

Collectively, the contributions in this Research Topic reflect a converging direction for the field: Multimodal integration increasingly requires methods that bridge biological scales, learn informative cross-modal representations under real-world constraints, and deliver robust, generalizable inference. Continued progress will benefit from shared benchmarks, transparent validation practices, and models that balance predictive performance with interpretability. This, in turn, would support reproducible multimodal neuroscience and facilitate the development of clinical applications and precision approaches, to improve prediction of outcomes and treatment response.

JS: Writing – original draft, Writing – review & editing. MN: Writing – review & editing. AC: Writing – original draft, Writing – review & editing.

The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The author AC declared that they were an editorial board member of Frontiers, at the time of submission. This had no impact on the peer review process and the final decision.

The author(s) declared that generative AI was used in the creation of this manuscript. During the preparation of this work the author(s) used ChatGPT 5.1 and GEMINI Flash 2.5 in order to improve the readability and language of the manuscript. After using these tools/services, the author(s) reviewed and edited the content as needed and take full responsibility for the content of the published article.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

AmuntsK.ZillesK. (2015). Architectonic mapping of the human brain beyond Brodmann. Neuron8, 1086–1107. doi: 10.1016/j.neuron.2015.12.001

BogdanR.SalmeronB. J.CareyC. E.AgrawalA.CalhounV. D.GaravanH.et al. (2017). Imaging genetics and genomics in psychiatry: a critical review of progress and potential. Biol. Psychiatry82, 165–175. doi: 10.1016/j.biopsych.2016.12.030

HeB.LiuZ. (2008). Multimodal functional neuroimaging: integrating functional MRI and EEG/MEG. IEEE Rev. Biomed. Eng.1, 23–40. doi: 10.1109/RBME.2008.2008233

HoneyC. J.SpornsO.CammounL.GigandetX.ThiranJ. P.MeuliR.et al. (2009). Predicting human resting-state functional connectivity from structural connectivity. Proc. Natl. Acad. Sci. U. S. A.106, 2035–2040. doi: 10.1073/pnas.0811168106

HorwitzB.PoeppelD. (2002). How can EEG/MEG and fMRI/PET data be combined?Hum. Brain Mapp.17:1. doi: 10.1002/hbm.10057

LiuJ.CalhounV. D. (2014). A review of multivariate analyses in imaging genetics. Front. Neuroinformat.8:29. doi: 10.3389/fninf.2014.00029

PhamK. T.SheyninJ.LiberzonI. (2024). “A blueprint for the next generation of clinical trials for precision psychiatry,” in Comprehensive Precision Medicine, ed. K. S. Ramos (Amsterdam: Elsevier), V1-83–V1-91. doi: 10.1016/B978-0-12-824010-6.00073-3

Trujillo-BarretoN. J.DaunizeauJ.LaufsH.FristonK. J. (2023). “EEG-fMRI information fusion: biophysics and data analysis,” in EEG-fMRI: Physiological Basis, Technique, and Applications, eds. C. Mulert, and L. Lemieux (Cham: Springer International Publishing), 695–726. doi: 10.1007/978-3-031-07121-8_28

ZhangY. D.DongZ.WangS. H.YuX.YaoX.ZhouQ.et al. (2020). Advances in multimodal data fusion in neuroimaging: overview, challenges, and novel orientation. Inform. Fus.64, 149–187. doi: 10.1016/j.inffus.2020.07.006

Keywords

deep learning, artificial intelligence, neuroimaging, machine learning (ML), microscopy, multimodal data fusion, Magnetic Resonance Imaging (MRI), precision medicine

Citation

Sheynin J, Naseri M and Callara AL (2026) Editorial: Multimodal brain data integration and computational modeling. Front. Neuroinform. 20:1838147. doi: 10.3389/fninf.2026.1838147

Received

24 March 2026

Revised

24 March 2026

Accepted

02 April 2026

Published

11 May 2026

Volume

20 - 2026

Edited and reviewed by

Michael Denker, Jülich Research Centre, Germany

Updates

Copyright

© 2026 Sheynin, Naseri and Callara.

This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

*Correspondence: Alejandro Luis Callara, alejandrocallara@gmail.com

Disclaimer

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.