Multimodal AI in Pediatric Neuroimaging

The interpretation of pediatric neuroimaging is uniquely challenging. A structural MRI, for instance, may show subtle abnormalities, but their clinical significance can be ambiguous without understanding the child's seizure patterns (from EEG) or genetic predispositions. Clinicians must constantly bridge these informational silos, but this mental synthesis is complex and imperfect. Multimodal AI offers a powerful solution by computationally fusing these data streams. By learning the intricate relationships between imaging features and non-imaging data, these models can contextualize findings, uncover hidden biomarkers, and unlock the full diagnostic potential of neuroimaging, moving beyond what single-modality analysis can achieve.This Research Topic aims to catalyze advances in AI that integrate pediatric neuroimaging with other data modalities for a deeper, more clinically actionable understanding of neurological disorders. Our goal is to move the field from siloed image analysis to integrated, patient-centered diagnostics.To achieve this, we welcome manuscripts focusing on, but not limited to, the following themes:•Showcasing AI models that combine neuroimaging (MRI, CT, etc.) with at least one additional data modality (e.g., EEG, genetics, clinical scores) for improved diagnosis, prognosis, or treatment planning.•Promoting the development of advanced AI architectures—such as data fusion techniques, graph-based models, or ensemble learning—explicitly tailored for multimodal integration.•Highlighting research that demonstrates how a multimodal approach yields synergistic insights, enhancing the value and clinical relevance of primary neuroimaging data.We invite a broad spectrum of submissions, including original research, reviews, and method papers, that feature neuroimaging as a core component of a multimodal AI strategy. Interdisciplinary collaborations that unite imaging experts with data scientists and clinical specialists are strongly encouraged. We are particularly interested in work that integrates neuroimaging with other data streams—such as electrophysiology (EEG), genomics, or clinical outcomes—to create a more holistic patient view.Crucially, all manuscripts must clearly demonstrate how the proposed multimodal approach provides a significant, quantifiable advantage over traditional, imaging-only analysis. Topic editors: Young-Tak Kim, Seho Lee Source: https://www.frontiersin.org/research-topics/72404/multimodal-ai-in-pediatric-neuroimaging