Introducing SHERPA

July 15, 2025

In January 2025, the Smart Human-centred Effortless support for Professional clinical Applications (SHERPA) project kicked off with a two-day meeting at the Philips Innovation Center in Best, Netherlands. SHERPA was awarded a €13.6 million grant by the Innovative Health Initiative and is set to innovate minimally invasive treatments in neurology and oncology. CIRSE is a partner in this important project as a task leader for education and training activities, and a junior partner in the dissemination, communication, and exploitation work package. SHERPA aims to support interventional radiologists by providing AI-powered assistive technologies as a seamless and trusted companion across workflows in neurology and oncology. Work package three will focus on “Intelligent Workflow Assistance for Oncological Interventions”, aiming to provide clinical evidence for the benefits of CBCT-guided ablations compared to standard care.

We spoke to one of CIRSE’s key opinion leaders in this project, Dr. Maarten Smits, about his expectations for this initiative, and about which aspects he is particularly excited about. Dr. Smits is an interventional radiologist and Program Director of Interventional Radiology at UMC Utrecht.

CIRSE Insider: SHERPA is all about AI-supported assistive technologies for IR workflows. How is AI supporting workflows in your current practice?

Smits: AI is indeed an important part of SHERPA, but the scope of the project is broader than just AI. Any innovations that can improve IR workflows are on the table such as smart navigation with cone-beam CT (CBCT), intra-arterial contrast injection, and robotics. At this moment, the role of AI in most IR practices is still very limited. There are several aspects in IR that could benefit from AI such as image post-processing, segmentation, and planning. We already use tools that assist with target segmentation, vessel analysis, and trajectory planning. However, real-time intraoperative AI support is still in its early stages. SHERPA aims to close that gap, making AI not just a planning aid, but an integral part of the procedure itself.

CIRSE Insider: Where do you see the biggest potential for AI in your daily practice?

Smits: In the short term, I see the biggest potential in image quality improvement. Any gain in image quality can directly result in either a lower radiation dose to the patient and personnel, or improved visualization of target structures. This is especially useful for CBCT, since these images are notoriously noisy and rich in artifacts. Improving the image quality of CBCT can help IRs start using this undervalued imaging modality.
In the longer term, there is a lot to gain in decision support and automation during image-guided procedures. Liver ablations, for instance, demand precise needle placement under challenging anatomical and procedural conditions. AI can streamline trajectory planning, automate alignment, and provide real-time guidance, thereby reducing the cognitive load on the operator and enhancing consistency across cases. Furthermore, AI-driven margin assessment tools could transform how we evaluate ablation success, enabling real-time feedback and the possibility to re-intervene within the same session if margins are insufficient.

CIRSE Insider: On a larger scale, where do you see AI improve interventional oncology treatments in the future?

Smits: AI will increasingly personalize and optimize treatment planning in interventional oncology. From identifying early-stage lesions suitable for ablation to predicting outcomes based on multi-modal data (imaging, genomics, clinical history), AI can tailor treatment strategies more precisely. AI will help the access to complex procedures by supporting less experienced practitioners with intelligent guidance.

CIRSE Insider: What do you think will change with the innovations that SHERPA is bound to bring?

Smits: The beauty of SHERPA is that it is a multinational consortium with a range of private and public parties involved, focussing solely on IR and improving workflows. Many projects in our field focus on improving outcomes, which is very important, but we have to face the global healthcare challenges such as an ageing population, shortages in (adequately trained) staff, and shortage in resources. Therefore, it is very important to look at how we can improve workflows to meet these challenges. For instance, the HepACAGA technique for liver ablations has already demonstrated benefit in terms of patient outcomes. Within this project we are investigating how aspects of this technique such as the single-room occupation, ease of needle planning, and the very short learning curve influence the workflow and these can be improved.

CIRSE Insider: Can you share with us which aspects of this project you are particularly excited about?

Smits: What excites me most is the convergence of technologies—robotics, AI, and advanced imaging—into a single workflow. The idea that we can use real-time image processing to automatically align a robotic device to a planned trajectory, is a major leap forward. I’m also enthusiastic about the development of tools integrated in the imaging modality that can assess ablation margins intra-procedurally, something we’ve long needed.
Finally, I would like to say that I am very glad to have CIRSE as a partner in this consortium. CIRSE has helped in determining meaningful objectives and will be instrumental in reaching out to the IR community for evaluation and dissemination of project outcomes.