Advancements in radiomics, machine learning, and deep learning have the potential to revolutionize cancer diagnosis and treatment by enabling more precise tumor differentiation, optimizing treatment planning, and facilitating early prediction of treatment outcomes. These AI-driven techniques leverage high-dimensional imaging features and patterns that are imperceptible to the human eye, enhancing tumor characterization, risk assessment, and therapy monitoring. Despite their promise, several challenges hinder clinical adoption, including the lack of automation in model implementation, segmentation bias, and the complexity of multi-step image pre-processing. These barriers hinder seamless integration into clinical workflows and limit the scalability of AI models for real-world applications.
This research topic aims to gather examples of state-of-the-art methods for robust, automated AI and quantitative MRI analysis of cancers across various organs, specifically focusing on the brain, breast, and prostate. Additionally, it seeks to explore challenges, pitfalls, and opportunities associated with translating AI innovations into clinically viable solutions.
This research topic aims to gather and advance knowledge on AI and deep learning models in cancer diagnosis and treatment. Specifically, it aims to highlight innovations in analysis techniques, model generalizability, and their impact on clinical outcomes.
Despite significant advancements in AI-driven radiomics and deep learning models, their clinical adoption remains limited due to challenges such as prolonged pre-processing times, variability in results, and limited practicality for routine use. To address these barriers, this research topic welcomes studies on recent advancements in AI-based quantitative MRI imaging methods that improve the clinical feasibility of radiomics and machine learning models for cancer diagnosis and treatment. A key focus is to improve the generalizability of AI frameworks such as use of automated segmentation, self-supervised learning, and other novel techniques. Contributions covering novel AI methodologies, validation studies, and translational applications that bridge the gap between research-driven AI models and real-world clinical practice are highly encouraged.
We welcome a range of article types, including Original Research and Literature Reviews, that explore key areas such as AI explainability, bias mitigation, generalizability, multimodal approaches, and the validation of AI-driven tools. Manuscripts addressing the following areas are of particular interest:
• AI- Driven Cancer Diagnostics and Treatment: Radiomics, habitat analysis, machine learning (ML), and deep learning (DL) methods that tackle clinical challenges, including early screening, diagnosis, tumor grading, prediction of molecular expression, treatment response assessment, and longitudinal risk and survival monitoring.
• Multimodal and multi-sequence Imaging Approaches: Development and application of radiomics, ML, and DL tools that integrate multiple imaging modalities for enhanced cancer characterization.
• Clinical Translation of AI Models: AI-driven radiomics, ML, and DL tools that have been successfully implemented in clinical practice.
• Expert Perspectives: Editorials discussing the advancements, challenges, and future directions of AI-driven radiomics and ML/DL in oncology.
Note: Studies lacking validation, either through internal or external cohorts or in vitro or in vivo, are outside the scope of this research topic and will not be considered for publication.
Keywords: MRI, Functional imaging, Radiomics, Habitat analysis, Auto-segmentation, Diagnosis, prognosis, personalized treatment
Important note: All contributions to this Research Topic must be within the scope of the section and journal to which they are submitted, as defined in their mission statements. Frontiers reserves the right to guide an out-of-scope manuscript to a more suitable section or journal at any stage of peer review.
