Nuclear medicine is a rapidly evolving domain at the intersection of advanced radiopharmaceuticals, hybrid imaging modalities such as PET/CT and PET/MR, and quantitative image analysis. Recent technological leapfrogs have brought forth unprecedented volumes of heterogeneous data, spanning dynamic imaging, clinical records, genomics, and more. However, this wealth of multimodal information is fraught with challenges: variability in acquisition protocols, limited labeled datasets for rare applications, domain shifts between institutions, and the pressing need for reproducible, interpretable, and clinically relevant workflows. While AI has already begun revolutionizing quantitative image reconstruction and lesion detection, ongoing debates persist regarding the harmonization of data, generalization across sites, and maintaining transparency amid increasingly complex algorithms. Notably, many recent studies demonstrate the power of deep learning—especially radiomics-informed and foundation models—but also highlight critical gaps in robust validation, generalizability, and regulatory adaptation. Effective clinical translation thus requires not just methodological innovation, but contextual awareness of bias, privacy, reproducibility, and operational constraints in nuclear medicine practice.

This Research Topic aims to consolidate pioneering contributions that harness artificial intelligence and multimodal data integration to advance the field of precision nuclear medicine from the research bench to clinical bedside. By encouraging studies that bridge technical breakthroughs with tangible clinical applications, the Research Topic endeavors to overcome persistent barriers such as scanner heterogeneity and rare tracer use. The ultimate objective is to foster reproducible, generalizable AI tools that elevate quantitative imaging, individualized decision support, and theranostic strategies, paving the way for truly personalized care in nuclear medicine. Key questions include: How can multimodal fusion enhance patient stratification and outcome prediction? What strategies ensure harmonization and trustworthiness of AI models? What are practical pathways for robust clinical deployment?

To gather further insights in integrating AI and multimodal data for clinically relevant nuclear medicine, we welcome articles addressing, but not limited to, the following themes:

- AI methods for PET/SPECT quantification, reconstruction, denoising, motion correction, segmentation, and lesion detection

- Multimodal fusion with CT/MRI, radiomics, pathology, omics, laboratory values, and EHRs for improved prognosis and treatment response prediction

- AI-driven theranostics and personalized dosimetry, including patient selection, dose planning, and longitudinal monitoring

- Strategies for harmonization, domain adaptation, uncertainty quantification, federated learning, and privacy-preserving data sharing across sites

- Applications and methodological advancements in foundation models, self-/weakly-supervised deep learning, and rare tracer scenarios

- Interpretability, reporting guidelines, fairness, workflow integration, and considerations for regulatory and clinical deployment.

We invite Original Research, Methods, Reviews, Mini-Reviews, Perspectives, Case Reports, and Clinical/Translational Studies. Authors are encouraged to ensure robust clinical context, transparent baselines, and, where feasible, validation in single-center or multi-center settings.

Are you interested in contributing a manuscript to this Research Topic?

Artificial Intelligence and Multimodal Data Integration for Advancing Precision Nuclear Medicine