Advances in Multi-Modal Artificial Intelligence for Oncology

Since its inception, artificial intelligence (AI) has been developed to mimic the adaptation and self-organization of living organisms or biological structures for finding novel solutions and making decisions within a data-driven framework [1]. It is therefore an evolutionary next step for AI to combine multiple sources of data for medical applications, mirroring the complex decision-making process of an oncologist [2]. These novel data integration methods may led to significant advancements in information processing techniques for clinical decision-support systems, generating unprecedented momentum in computational oncology and precision medicine.In particular, multi-modal AI has emerged as a key enabler for integrating heterogeneous oncology data, including medical imaging, genomics, transcriptomics, radiomics, deep features, and clinical parameters [3]. Deep learning models or radiomic features combined with other -omics can potentially exemplify this paradigm by establishing associations between radiological features and molecular expression to uncover the underlying disease mechanisms and support personalized diagnostics. In parallel, multi-omics research has demonstrated the potential to holistically capture intra-tumor variability, improving prediction performance, model stability, and prognostic accuracy [4]. However, multi-modal integration remains challenged by complexities in infrastructure, data sharing legislation, incomplete data points, and missing clinical variables [5]. Consequently, generative models [6], explainable AI [7], and federated learning [8] have become essential components for developing robust, interpretable, and privacy-preserving multi-modal AI systems in oncology.The aim of this special issue is to highlight recent advances and novel applications of multi-modal artificial intelligence in oncology, with a particular focus on deep learning-based methodologies that integrate heterogeneous data sources. Multi-modal AI enables the fusion of medical imaging, genomics, transcriptomics, radiomics, and clinical data into unified feature spaces, supporting more precise and robust decision-support systems for cancer diagnosis, prognosis, and treatment planning. Despite significant progress in multi-modal imaging technologies and computational methods, a critical knowledge gap persists between tissue-level imaging phenotypes and the underlying molecular and genetic disease biomarkers.Integrating complementary data modalities, such as imaging, clinical, laboratory, semantic, -omics and radiomics, into composite signatures has the potential to alleviate the limitations of single-source data by capturing biological pathways, clinically relevant knowledge, and tumor heterogeneity. Such holistic representations has the potential of improving therapy response assessment, survival analysis, and prognostic modeling. Furthermore, incorporating a wider range of semantic and contextual features may lead to human-centric AI models, enhancing interpretability, trustworthiness, and clinical acceptance while advancing personalized oncology care.In this Special Issue, we welcome novel methodological contributions, benchmarking and comparative studies of multi-modal AI approaches, critical review articles, methodology surveys focusing on data integration strategies, as well as the introduction of new multi-modal datasets and high-quality translational studies relevant to oncology. Contributions addressing theoretical foundations, algorithmic developments, clinical validation, and real-world deployment challenges of multi-modal artificial intelligence are particularly encouraged. Both exploratory and application-driven studies that advance the state of the art in cancer diagnosis, prognosis, and treatment planning are within the scope of this Special Issue. In particular, indicative topics of interest include, but are not limited to:• From Imaging to Genomics: Radiogenomics, Radiotranscriptomics, Radiocytogenetics in Oncology Artificial Intelligence• Generative Artificial Intelligence in Multi-Modal Oncology Datasets• Multi-Modal Federated Learning in Oncology• Explainable Artificial Intelligence in Multi-Modal Oncology Artificial Intelligence• Agentic Artificial Intelligence for Multi-Modal Cancer Diagnosis• Multi-Expert Artificial Intelligence for Prognostic Oncology• Prognostic Multi-Modal Artificial Intelligence in Medicine• Multi-Modal Artificial Intelligence in Cancer Diagnosis• Multi-Omic Integration of Oncology Data with Machine Learning• Ensemble Artificial intelligence for Multi-Parametric Medical Imaging in Cancer Diagnosis• Multi-Modal Data Fusion for Capturing the Complex Underlying Biological Mechanisms• Multi-Omic Deep Learning in Cancer Prognosis and Diagnosis• Multi-Source Data Integrations of Radiomics, Genomics, Laboratory Examinations, and Clinical dataPlease note that manuscripts consisting solely of bioinformatics or computational analysis of public omics databases that are not supplemented by relevant functional validation (clinical cohort or biological validation in vitro or in vivo) are out of scope for this Research Topic.References[1] Trivizakis, E., & Marias, K. (2023). Deep Learning Fundamentals. In Introduction to Artificial Intelligence (pp. 101-131). Cham: Springer International Publishing. DOI: 10.1007/978-3-031-25928-9_6[2] Trivizakis, E., Koutroumpa, N. M., Souglakos, J., Karantanas, A., Zervakis, M., & Marias, K. (2023). Radiotranscriptomics of non-small cell lung carcinoma for assessing high-level clinical outcomes using a machine learning-derived multi-modal signature. BioMedical Engineering OnLine, 22(1), 125. DOI: 10.1186/s12938-023-01190-z[3] Trivizakis, E., Koutoulakis, E., Koutoulidis, V., Moulopoulos, L. A., Terpos, E., Ntanasis-Stathopoulos, I., ... & Marias, K. (2025, May). Radiocytogenetics in Multiple Myeloma: Predicting Cytogenetic Aberrations from WBCT Imaging Features. In 2025 IEEE 19th International Symposium on Applied Computational Intelligence and Informatics (SACI) (pp. 000231-000236). IEEE. DOI: 10.1109/SACI66288.2025.11030160[4] Trivizakis, E., Souglakos, J., Karantanas, A., & Marias, K. (2021). Deep radiotranscriptomics of non-small cell lung carcinoma for assessing molecular and histology subtypes with a data-driven analysis. Diagnostics, 11(12), 2383. DOI: 10.3390/diagnostics11122383[5] Trivizakis, E., Papadakis, G. Z., Souglakos, I., Papanikolaou, N., Koumakis, L., Spandidos, D. A., ... & Marias, K. (2020). Artificial intelligence radiogenomics for advancing precision and effectiveness in oncologic care. International journal of oncology, 57(1), 43-53. DOI: 10.3892/ijo.2020.5063[6] Dimitriadis, A., Trivizakis, E., Papanikolaou, N., Tsiknakis, M., & Marias, K. (2022). Enhancing cancer differentiation with synthetic MRI examinations via generative models: a systematic review. Insights into Imaging, 13(1), 188. DOI: 10.1186/s13244-022-01315-3[7] Koutoulakis, E., Trivizakis, E., Markodimitrakis, E., Agelaki, S., Tsiknakis, M., & Marias, K. (2026). A critical review of explainable deep learning in lung cancer diagnosis. Artificial Intelligence Review, 59(1), 28. DOI: 10.1007/s10462-025-11445-x[8] Markodimitrakis, E., Trivizakis, E., Ioannidis, G. S., Koutoulakis, E., Tsiknakis, M., Papanikolaou, N., ... & Marias, K. (2026). Federated learning on magnetic resonance imaging: a critical review. Artificial Intelligence Review. DOI: 10.1007/s10462-026-11508-7 Topic editors: Eleftherios Trivizakis, Mohammad Masum, Satishkumar Chavan Source: https://www.frontiersin.org/research-topics/79236/advances-in-multi-modal-artificial-intelligence-for-oncology