The first domain the authors address is artificial intelligence, and they deliberately reject the framing that has dominated the conversation. The question, they insist, is not whether AI will replace the interventional oncologist, but which interventional oncologists will master AI and use it to redefine the standard of care. The applications are no longer speculative. AI-assisted needle tracking, intraprocedural image registration and automated target confirmation are moving from research prototypes into clinical tools. Automated tumour contouring and ablation margin prediction are reducing the cognitive burden of treatment planning while improving reproducibility across operators and institutions. Perhaps most transformative is radiomics-based early response assessment, which promises to detect treatment failure before conventional imaging shows any change, a paradigm shift for follow-up and retreatment decisions that builds on the foundational radiomics work published in Nature Communications by Aerts and colleagues. Beyond the procedure itself, AI is beginning to attack the administrative and workflow inefficiencies that erode clinical capacity, from scheduling optimisation to structured reporting. The cumulative effect, the authors argue, is a reduction in cognitive load that frees the physician to focus on what remains irreplaceable: clinical judgement, the patient relationship and procedural expertise. Their conclusion is blunt and likely to circulate widely: those who master these tools will not be replaced by AI; they will replace those who did not.

The second domain is the physical environment in which image-guided cancer therapy is delivered. Interventional radiology was born from the insight that imaging and therapy are not sequential but simultaneous, a principle traceable to Dotter and Judkins’ description of transluminal angioplasty in 1964 and to Seldinger’s landmark catheter technique a decade earlier. Gangi’s own group helped define the hybrid era in the early 1990s, when they first described combining computed tomography with C-arm fluoroscopy, uniting cross-sectional precision with real-time dynamic guidance in a single interventional environment. That innovation proved its worth rapidly in one of the most clinically urgent corners of the field: cancer pain management. CT-guided procedures ranging from coeliac plexus neurolysis to vertebroplasty established early that the specialty could address not only tumour burden but the suffering it causes, an integration of palliation within an oncological framework that tumour boards, the authors note, have been slow to fully recognise. Three decades of progressive integration have since brought ultrasound, interventional MRI, PET-CT fusion planning and real-time computational overlay into environments no other specialty has claimed as its own. The vision for 2030 is the fully integrated CT, angiography and MR suite as a standard of care in academic centres, with interventional MRI transitioning from a specialised capability to a broadly implemented clinical platform. Gangi’s institution has treated endometriosis, renal tumours, uterine adenomyosis and bone metastases under real-time MR guidance for nearly two decades. What was exceptional in 2007, the authors contend, must become routine by 2030.

The third domain is evidence, and here the lecture delivers its harshest self-assessment. Interventional oncology has earned its seat at the tumour board, but that seat was built case by case through outcomes data that could no longer be ignored. Thermal ablation, transarterial chemoembolisation, radioembolisation and vertebroplasty have each demonstrated clinical value, supported by standardisation efforts such as the terminology and reporting criteria established by Goldberg and colleagues in Radiology and the clinical validation recommendations for liver tumour ablation advanced by Lencioni, de Baère and colleagues. Yet presence at the table is not influence at the table, and the authors draw a sharp distinction between the two. The currency at a multidisciplinary meeting is data, and the persistent absence of phase III randomised controlled trials in interventional oncology is identified as the specialty’s most significant vulnerability. Tumour boards respect what they can systematically evaluate, and excellent results must be matched by an equally excellent evidence architecture. The challenge for the coming decade is to move from the retrospective series that formed the foundation of the field toward prospective, randomised and ideally multicentre trials.

Importantly, the authors do not treat randomised trials as the only legitimate form of evidence. They argue that prospective registries, pragmatic multicentre studies and robust real-world data analyses should be regarded as integral components of a mature evidence ecosystem, capturing patient heterogeneity, treatment sequencing, institutional variation and the rapid evolution of devices in ways conventional trials cannot always accommodate. These approaches are not a lower level of ambition but a complement that generates clinically meaningful data for multidisciplinary decision-making and guides the design of future randomised studies. Several frontier areas demand particular attention. The synergy between thermal ablation and immunotherapy, including the abscopal effect as a therapeutic target, is generating prospective data that could transform interventional oncology’s role in systemic cancer treatment, with trials such as the combination of tremelimumab and ablation in advanced hepatocellular carcinoma pointing the way. Radiomics-based biomarkers are moving toward regulatory approval. Theranostics and radiopharmaceutical combinations represent an emerging paradigm. And the outpatient ablation centre model, delivering image-guided therapy ambulatorily at lower cost and with reduced institutional dependency, may prove to be the most disruptive development of the decade.

The fourth domain is the most personal: the relationship between the interventional oncologist and the patient. The authors pose a question they have asked audiences across Europe, and the answer is always the same. Ask a patient who performed their surgery and they name the surgeon. Ask who performed their biopsy or their ablation and they name the building. This asymmetry, they argue, is not inevitable but a consequence of how the specialty has historically organised itself around a technician model in which the patient is referred, the procedure is performed and the patient is returned. That model is both clinically suboptimal and institutionally fragile, because it makes the practitioner replaceable. The physician model is their prescription for the future: dedicated outpatient consultations, a full therapeutic relationship with continuity of care and structured follow-up, inpatient management when required, and accreditation through the International Accreditation System for Interventional Oncology Services as institutional validation of the standard. When a patient has a named interventional oncologist, someone they can call, someone who reviews their imaging and explains a recurrence, the specialty becomes indispensable. The authors are explicit that this is not a turf battle in the pejorative sense but a claim to clinical ownership that outcomes justify, and they note that their doors remain open to surgeons who recognise the value of minimally invasive, image-guided therapy. The transformation from proceduralist to physician, they write, is not a threat to the specialty’s identity but its completion.

The fifth domain is mentorship, and it is here that the tribute to Thierry de Baère becomes most direct. The challenges of 2030 will be met by those being shaped today, and de Baère understood this better than most. He did not simply transmit techniques; he transmitted a philosophy of rigour, intellectual honesty and care that extended beyond the procedure, building careers with the same intentionality he brought to clinical innovation. The authors describe him as an architect of people. The mentor’s role, they argue, is not to produce competent operators but to produce the next generation of leaders, physicians who can navigate a multidisciplinary tumour board, design a prospective trial and inspire a trainee as they themselves were once inspired. Resilience, ethics and clinical empathy cannot be learned from a textbook; they are modelled in the culture of a department, in how a senior colleague responds to a complication, and in whether a trainee is protected or exposed when things go wrong. Professional bodies including CIRSE and IASIOS are working toward harmonised, competency-based training curricula, building on standards of practice such as those published by Crocetti, de Baère and colleagues on thermal ablation of liver tumours, and the EBIR and EBIR-IO certification pathways are meaningful steps. But certification, the authors caution, is a floor, not a ceiling. What the specialty owes the next generation is not compliance with a minimum standard but the ambition to exceed it.

The trajectory the lecture traces is remarkable when viewed as a whole. From Seldinger’s needle and Dotter’s catheter, through the CT-fluoroscopy hybrids of the 1990s and the MR-guided suites of the 2000s, interventional oncology has built a specialty that is less invasive, less expensive, less disabling and less time-consuming than the alternatives it has progressively replaced. The authors frame the field’s past, present and future as a continuous arc of technical courage, clinical ambition and intellectual generosity, one that aligns closely with the legacy of Thierry de Baère, whose work and mentorship helped establish precisely these foundations. The vision they articulate is of a multidisciplinary, evidence-driven, scientifically rigorous and clinically integrated discipline, recognised as the fourth pillar of oncological treatment alongside surgery, medical oncology and radiation therapy.

The work ahead, as the authors summarise it, is a five-point charge to the specialty: master AI before it masters the agenda; own the multimodal suite before others claim it by default; lead the multidisciplinary team with data rather than deference; claim the patient as a physician rather than a technician; and build the future by investing unconditionally in those who will carry it forward. It is a message aimed squarely at trainees and early-career physicians, but its implications reach hospital administrators, tumour boards and device developers alike. The authors close with a reminder that carries the urgency of a countdown: 2030 is closer than we think, and it begins today.

Subject of Research: The future of interventional oncology, including artificial intelligence, hybrid imaging suites, evidence development, clinical ownership and mentorship, as outlined in the inaugural Thierry de Baère Lecture.

Article Title: Interventional oncology 2030: challenges and opportunities for the next generation — in honour of Thierry de Baère

Article References: Gangi, A., & Palussière, J. (2026). Interventional oncology 2030: challenges and opportunities for the next generation — in honour of Thierry de Baère. CVIR Oncology, 2(1), Article 18. https://doi.org/10.1007/s44343-026-00048-0

Image Credits: AI Generated

DOI: 10.1007/s44343-026-00048-0

Keywords: interventional oncology, artificial intelligence, thermal ablation, hybrid imaging suites, radiomics, immunotherapy, abscopal effect, multidisciplinary tumour boards, IASIOS accreditation, mentorship, Thierry de Baère, randomised controlled trials