Radiation oncology is undergoing a profound evolution driven by advances in imaging, artificial intelligence (AI), and precision treatment planning. These innovations aim to increase treatment efficacy, reduce side effects, accelerate planning and delivery, and personalize therapy according to patient-specific anatomy and tumor biology.
Regulatory imperatives, clinical demand, and technological maturity converge to shape a new era in which imaging and AI are integrated into radiation delivery workflows. This blog reviews the state of the art in imaging and AI in radiation oncology, explores precision therapy methods (including adaptive radiation), presents emerging case studies, and discusses implementation and regulatory challenges.
According to Kings Research, the global radiation oncology market was valued at USD 8.57 billion in 2023 and is projected to expand from USD 8.99 billion in 2024 to USD 13.16 billion by 2031, reflecting a 5.60% CAGR over the forecast period. This steady growth reflects rising cancer prevalence and the adoption of AI-driven imaging and adaptive radiotherapy, moving the field toward more precise and personalized care.
How imaging, data, and AI form the foundation of modern radiation oncology
Imaging Modalities and Data Sources:
Advances in imaging technology underpin precision therapy. CT, cone-beam CT (CBCT), MRI, PET, and hybrid modalities generate high-resolution anatomical, functional, and molecular information. Imaging informs tumor delineation (target volumes), normal tissue segmentation, motion management, and treatment verification. Multiple daily imaging (e.g., CBCT) enables assessment of inter-fractional changes in patient positioning, tumor motion, or deformation.
Data sources in radiation oncology include electronic medical records, diagnostic imaging archives, radiotherapy planning systems, record-and-verify systems, treatment delivery data, and outcomes (patient follow-up). These heterogeneous datasets are essential inputs to AI models seeking prediction, automation, or quality control. Reviews in the literature affirm that radiation oncology already has a strong dose of historical imaging and planning data suited to AI applications. (Source: pmc.ncbi.nlm.nih.gov)
AI Use-Cases in Oncology: From Detection to Decision Support:
AI has been applied in radiation oncology for auto-segmentation of organs and tumors, dose prediction, plan optimization, image enhancement, motion modeling, and quality assurance. A recent NIH study demonstrated an AI tool that uses routine clinical data to predict how patients will respond to certain cancer therapies, thus contributing to treatment selection. (Source: www.nih.gov)
Another example is collaborations between software providers to build AI-enabled treatment planning tools. Elekta and MIM Software (acquired by GE Healthcare) announced a strategic collaboration to combine auto-contouring, treatment planning, and workflow automation. The goal aims at increasing throughput, improving precision, and reducing planning time.
Safe and ethical deployment of AI demands governance. The European Union’s AI Act now applies to high-risk AI systems in healthcare. The regulation mandates data governance, traceability, transparency, human oversight, and risk management.
Precision Therapy: Adaptive Radiation and Workflow Innovations
Adaptive Radiation Therapy (ART):
Adaptive radiation therapy adjusts treatment in response to changes in patient anatomy, tumor size, or position over the course of therapy. Daily imaging (e.g., CBCT) allows detection of deviations in tumor location or shape. Software tools can alter treatment plans accordingly, reducing dose to healthy tissue while maintaining or escalating tumor dose. Elekta has introduced innovations in CBCT acquisition and reconstruction (scatter correction, iterative reconstruction, and “polyquant” modeling) to improve image quality quickly and enable more practical adaptive planning.
ART reduces uncertainties inherent in static planning margins. For example, bladder or bowel filling can shift the tumor position. Adaptive workflows shrink planning margins or allow online modification of plan geometry. The result can be fewer side effects, a higher therapeutic ratio, or both.
Workflow Acceleration and Automation:
AI assists in automating the segmentation of organs or target volumes, which traditionally is labor-intensive. Treatment planning software enhancements, vendor collaborations, and cloud-based tools allow faster plan generation. Collaboration between Elekta and GE’s MIM Software aims to deliver vendor-agnostic, automated contouring and planning tools. The strategic partnership announced in April 2024 intends to reduce the time between imaging and plan delivery while maintaining or improving quality.
Quality assurance (QA) also benefits from AI: in detecting errors in plan geometry, dose calculation, or delivery deviations. AI models can cross-validate planned vs delivered doses using imaging feedback or log files.
AI and Adaptive Therapy in Radiation Oncology
- Elekta and GE Healthcare Collaboration: Elekta and GE Healthcare entered a global commercial collaboration to integrate imaging with precision radiation therapy. Their combined offering aims to enable simulation, guidance, and treatment delivery in a unified workflow, especially in regions where imaging infrastructure and radiotherapy are lacking or disparate. This collaboration supports more precise tumor targeting and spares healthy tissue.
- Sun Nuclear’s Acquisition of Oncospace: Sun Nuclear, part of Mirion Medical, acquired Oncospace, a software provider focused on AI-powered radiation oncology solutions. Oncospace developed tools for predictive plan quality, plan feasibility, and conversion of treatment planning archives. Following this acquisition in 2025, Sun Nuclear aims to expand its quality assurance digital-forward strategy and support clinics in optimizing treatment planning via AI assistance (Source: ir.mirion.com).
- Elekta’s Adaptive Radiotherapy Enhancements: Elekta has advanced its adaptive radiotherapy (ART) capabilities by refining CBCT imaging, reducing scatter artifact, improving iterative reconstruction, and integrating quantification models (e.g., polyquant) to better map tissue density. These enhancements are designed to make daily adaptation more practical by improving imaging speed, image accuracy, and reducing time overhead in the planning workflow.
The biggest challenges and regulatory hurdles for AI in radiation oncology
Image Quality, Uncertainty, and Safety:
Precision therapy depends on high-quality imaging. Imaging artifacts, motion blur, scatter in CBCT, or inaccuracies in deformable image registration degrade confidence. AI models may propagate errors or bias if trained on non-representative datasets. Real-time or near-real-time imaging also demands balancing speed with dose to the patient.
Regulatory bodies require validation and verification of AI tools. Data privacy, explainability, traceability, robustness, and human oversight are necessary in healthcare AI. Under the EU AI Act, AI systems designated high risk (including many diagnostic and therapeutic tools) must satisfy specific obligations.
How to integrate AI tools into clinical radiation oncology workflows:
Clinics need to integrate AI tools and imaging improvements into existing radiation oncology workflows. Staff (radiologists, radiation oncologists, medical physicists, dosimetrists) must develop trust, understand limitations, and verify outputs. Training in data science and AI literacy becomes essential: national cancer institutes and societies increasingly hold workshops to train medical physicists and oncologists in these domains.
Changing workflow to allow adaptive planning introduces scheduling, staffing, and quality assurance demands. Clinics must balance throughput and patient volume with the time required for daily imaging, plan adaptation, and verification.
Balancing data, privacy, and ethics in AI-powered radiation oncology:
Effective AI systems require large, annotated imaging and outcomes datasets. Privacy regulation (e.g., HIPAA in the U.S., GDPR in the EU) limits how data can be shared. Federated learning and privacy-preserving techniques represent promising models. An example is the Cancer AI Alliance (CAIA), a collaboration that uses federated AI learning frameworks, maintaining data within each center while aggregating results across centers. CAIA expects first insights by the end of 2025.
Ethical considerations include bias in datasets, transparency in AI decision support, oversight of automated decision suggestions, and ensuring that AI supports rather than replaces human judgment.
Future directions for AI imaging and adaptive therapy in radiation oncology
Emerging trends suggest several axes of progress. First, imaging modalities will continue to improve, including integration of molecular imaging, MRI-guided radiotherapy, PET guidance, and functional imaging that reflect tumor biology (hypoxia, metabolism). Second, AI will evolve toward predictive models that integrate imaging, genomics, treatment history, and patient-specific data to personalize treatment planning beyond anatomical contouring.
Third, adaptive and online therapy will spread: architectures that permit fast imaging, plan updating, and delivery in the same session (or near real-time) will become more feasible. Fourth, regulatory frameworks (AI regulation, medical device regulation) will tighten and evolve to provide clearer paths for approval, audit, safety, and post-market monitoring.
Conclusion
Radiation oncology is entering a phase where imaging, AI, and precision therapy converge to deliver more effective, safer, and faster treatments. Progress in adaptive radiotherapy, AI-enabled segmentation and planning, predictive modeling, and vendor collaborations (e.g., Elekta & GE Healthcare, Sun Nuclear & Oncospace) highlight tangible steps forward.
Challenges in image quality, regulatory compliance, workflow integration, and ethical use must be addressed. Successful implementation depends on combining technological innovation with robust validation, clinical acceptance, and regulatory alignment. The future of radiation therapy promises greater personalization, improved outcomes, and potentially lower toxicity for patients.
