In Brief | Interventional Radiology | Oncology | AI
The Expanding Role of Artificial Intelligence in Interventional Radiology
Time to read: 03:48
Time to listen: 07:12
Type of article: Narrative
MedED Catalogue Reference: MNR001
Category: Interventional Radiology
Cross Reference: Artificial Intelligence, Radiology
Keywords: AI, IR, Interventional Radiology, Oncology, ablation
Key Takeaway
This article is original. Any misstatements or misrepresentations are unintentional. Every effort has been made to reference sources where applicable. Any failure to references is unintentional and will be addressed appropriately. Access the full Disclaimer Information.
Listen to Full Article
Artificial intelligence (AI) is poised to revolutionize interventional radiology (IR), with the potential to transform every stage of patient care—from initial selection through to real-time procedural guidance and post-treatment follow-up. IR’s reliance on advanced imaging, precision techniques, and complex decision-making makes it especially well-suited for AI integration.
However, several challenges continue to limit widespread adoption. These include a lack of standardized, high-quality datasets, operator-specific variability, and inconsistency in imaging protocols. Furthermore, subjective decision-making, an absence of robust clinical guidelines, and limited AI exposure in current training programs contribute to a significant knowledge gap, particularly among practising clinicians.
Despite these obstacles, AI is already being implemented in a number of impactful ways across IR. Below are some of the current applications demonstrating its clinical utility
Patient Selection
Clinical applications—particularly in patient selection—highlight the real-world potential of AI in IR. This is especially evident in interventional oncology, where predictive models are already being used to stratify patients and optimise therapeutic strategies.
Advanced algorithms now assist in predicting tumour response to intra-arterial therapies by integrating radiological and clinical data. More recently, foundation models—deep learning architectures based on transformers—have emerged as a significant leap forward. These models are trained on large, multimodal datasets using self-supervised learning, reducing the need for expensive manual labelling. They enable generalisation across tasks, perform well with limited data (zero-shot learning), and can integrate diverse data inputs, including imaging and text.
Models like MedPaLM have shown promise in enhancing diagnostic accuracy, identifying multimodal biomarkers, and refining personalized treatment pathways—making them highly relevant to IR’s data-rich environment
AI-Driven Decision Support
AI-driven software offers significant advancements in enhancing procedural planning and execution in IR.
Examples of how AI is being implemented include:
Decision support tools that currently offer real-time optimisation of needle trajectories, as seen in AI-assisted CT-guided lung biopsies. These tools enable faster, safer procedures and may eventually integrate with robotic systems for automation.
In thermal ablation, AI-driven computational modelling helps tailor ablation zones, potentially improving local control rates.
Deep learning is also enhancing digital subtraction angiography (DSA) by generating synthetic images that reduce motion artefacts and radiation exposure.
AI-enhanced fluoroscopy platforms with ultrafast collimation offer another advancement, dynamically limiting radiation to the region of interest. This has been shown to reduce patient dose and scatter radiation to staff, improving overall procedural safety.
Live Assistance and Automatic Segmentation
Real-time AI support is increasingly valuable during procedures, particularly for tracking devices and preventing complications. DL models can automatically segment catheters and detect tips in cerebral angiography, improving device visualisation across multiple screens.
Advanced systems can now monitor wires, catheters, and embolic agents in real time, issuing alerts if devices stray from predefined zones or move off-screen. This assists in the early detection of non-target embolisation.
While vessel segmentation and lesion analysis have been extensively developed in interventional cardiology, their adoption in IR remains limited. Integrating such tools could significantly enhance procedural precision, especially in complex or high-risk cases.
AI and Robotics: A Converging Frontier
Robotic systems are rapidly evolving in IR, particularly in percutaneous and endovascular domains. In percutaneous interventions, robotics has shown sub-millimetre accuracy in thermal ablation of tumours, with the ability to replicate complex, multi-needle trajectories—capabilities often beyond human precision.
These systems reduce radiation exposure and inter-operator variability, making them especially valuable for less experienced clinicians. Looking ahead, AI integration into robotics—such as path planning and margin assessment—could pave the way for fully automated platforms.
Technologies such as simulation-to-reality (Sim2Real) and reinforcement learning enable robotic systems are already being used to train in high-fidelity, patient-specific simulations before transitioning to real-life applications, thereby enhancing safety and personalisation.
Robotic assistance is also gaining traction in endovascular procedures. However, limitations such as insufficient haptic feedback and collision detection remain barriers to full autonomy. Innovations like magnetorheological fluid resistance systems and haptic vision sensors are beginning to address these issues, enabling more intuitive and safer navigation within vascular systems.
Looking Ahead: Beyond Imaging
While AI’s current applications are largely image-centric, its potential in IR extends much further. AI can streamline clinical trials, personalise treatment pathways, and support longitudinal patient monitoring. Real-time intraoperative support, automated reporting, and predictive analytics are just the beginning.
For AI to realise its full impact in IR, robust collaboration between clinicians, data scientists, and engineers is crucial. This interdisciplinary approach will ensure that AI solutions are clinically relevant, safe, and seamlessly integrated into real-world practice.
Conclusion
Artificial intelligence is poised to redefine interventional radiology across every phase of patient care.
From sophisticated predictive models and intraoperative guidance to AI-robotic systems and beyond, the synergy between human expertise and machine intelligence holds the key to safer, more efficient, and more personalised interventions. As the field advances, the onus is on clinicians to engage with these tools, drive responsible innovation, and ensure AI becomes an asset, not a barrier, to patient-centred care.
Sources
- Lesaunier, A., Khlaut, J., Dancette, C., Tselikas, L., Bonnet, B., & Boeken, T. (2025). Artificial intelligence in interventional radiology: Current concepts and future trends. Diagnostic and interventional imaging, 106(1), 5–10. https://doi.org/10.1016/j.diii.2024.08.004
- Posa A, Barbieri P, Mazza G, et al. Technological advancements in interventional oncology. Diagnostics (Basel). 2023;13:228. doi: 10.3390/diagnostics13020228
- Zhang, H., Kulkarni, S., Jiao, Z., & Bai, H. X. (2024). Potential applications of AI within interventional oncology include patient diagnosis, response prediction, and intraprocedural guidance, dependent upon standardized and responsible implementation. Endovascular Today. Retrieved 13 May 2025. https://evtoday.com/articles/2023-oct/the-future-of-artificial-intelligence-in-interventional-oncology
Back to top
Disclaimer
This article is compiled from several resources researched and compiled by the contributor. It is in no way presented as an original work. Every effort has been made to attribute quotes and content correctly. Where possible all information has been independently verified. The Medical Education Network bears no responsibility for any inaccuracies which may occur from the use of third-party sources. If you have any queries regarding this article contact us
Fact-checking Policy
The Medical Education Network makes every effort to review and fact-check the articles used as source material in our summaries and original material. We have strict guidelines in relation to the publications we use as our source data, favouring peer-reviewed research wherever possible. Every effort is made to ensure that the information contained here is an accurate reflection of the original material. Should you find inaccuracies, out of date content or have any additional issues with our articles, please make use of the contact us form to notify us.
