ECR 2026 AI Landscape

The European Congress of Radiology (ECR) 2026 takes place in Vienna from March 4–8 under the theme “Rays of Knowledge.” The programme spans scientific research, subspecialty education, professional challenges, and multidisciplinary practice.

Artificial intelligence is a prominent thread across the agenda. Two dedicated strands are especially visible: the AI Theatre and the In Focus programme (“The Art of AI in Clinical Practice”). Beyond these, AI-tagged sessions appear across research formats, professional tracks, and clinical subspecialties. This report summarises how AI is distributed across the official programme structure and tagging.

Want the industry-side view of the same conference? See ECR 2026 AI Industry Landscape, our analysis of exhibitors, vendor mix, and geographic clustering. For launch activity and product positioning, see ECR 2026 AI Product Landscape.

Highlights at a glance

• 142 of 603 sessions are AI-tagged (23.5%).
• Most AI sessions appear in Research Presentation Sessions and ESR at Work Sessions.
• AI content splits into 53.5% clinical and 46.5% structural themes.
• Leading clinical areas: Oncologic Imaging and Neuro.
• 7 sessions reference generative AI.

For a quick comparison to the exhibitor floor (who is building and selling), cross-reference Highlights and Vendor mix.

Where AI shows up in the programme

AI visibility is not evenly distributed across formats¹. Session types designed for research presentation, structured discussion, and practice-oriented problem solving naturally attract AI work, because they provide a venue for reporting validation studies, deployment experiences, and governance questions.

More technique-focused or case-driven formats often include AI as a supporting tool rather than the organising theme. This is consistent with how adoption is playing out: evidence generation and operational integration remain the main drivers of AI’s programme footprint.

The chart below compares AI session counts against total sessions per format, showing where AI appears most frequently within the official programme structure.

On the industry side, these “evidence and integration” formats tend to correlate with a stronger presence of workflow and platform vendors. See AI density by vendor type and Vendor type by region.

Looking only at total counts can hide an important distinction: some formats are large and therefore accumulate many AI sessions, while other formats may be smaller but heavily AI-dense. The composition view below focuses on the most AI-active formats and shows the AI vs non-AI split within each.

Clinical versus structural focus inside AI

AI sessions divide between organ-system clinical content and structural themes such as workflow integration, informatics, governance, education, and professional adaptation. Clinical sessions tend to emphasise diagnostic performance and disease-specific application, while structural sessions focus on how AI is deployed, monitored, and embedded into daily operations.

The size of the structural share matters. It indicates that the conversation has moved beyond model performance alone and is increasingly shaped by scalability, accountability, and institutional readiness.

This split has a direct mirror on the exhibitor floor: as structural themes grow, “AI enabling” vendors (workflow, platforms, integration) become more visible. See AI density by vendor type and Vendor mix.

Clinical AI by subspecialty

Distribution across organ systems is uneven. AI activity tends to concentrate in high-volume domains and in pathways where imaging directly influences downstream management decisions. Oncology-linked imaging and cross-sectional workflows often attract more AI work because datasets are larger, outcomes are measurable, and reporting patterns are structured.

Lower-density subspecialties do not imply limited relevance. They may reflect data availability, regulatory complexity, or slower workflow transformation in those domains. Over time, diffusion into additional subspecialties is expected as tooling stabilises and evidence accumulates.

The chart below shows AI session counts by clinical subspecialty, based on the programme’s subspecialty tagging.

On the exhibitor side, modality positioning provides a complementary lens on demand. See Technology footprint.

Operational footprint

A large portion of the AI programme is operational in nature. Structural tags frequently co-occur with AI sessions, highlighting informatics architecture, integration into reporting systems, governance, quality assurance, and professional adaptation. The overlap between informatics and clinical sessions reinforces a practical reality: the bottleneck is often deployment, not model performance.

In total, 13 AI sessions sit at the intersection of imaging informatics and clinical practice.

Generative AI appears in 7 sessions. The relatively modest count suggests LLMs are currently positioned as complementary tools, likely focused on reporting support, communication, and workflow augmentation rather than replacing image interpretation.

The chart below shows the most common structural tags that appear within AI sessions. Counts are non-exclusive, so they reflect thematic presence rather than mutually exclusive categories.

The exhibitor mirror of this is the workflow and platform layer. Cross-reference Vendor type by region and AI density by vendor type.

ETC level distribution

The European Training Curriculum (ETC) levels provide a structured view of educational depth. Comparing the percentage distribution between AI sessions and the overall programme shows whether AI is concentrated in advanced content or embedded across general radiology education.

A skew toward higher levels would suggest AI remains largely within specialist or expert audiences. A flatter distribution, especially with meaningful representation at foundational levels, indicates that AI is being treated as a core competency rather than a niche interest.

The chart below compares the level distribution as percentages for AI sessions versus all sessions.

CME weight

CME allocation offers a practical proxy for formal educational emphasis. AI sessions total 173 CME credits across 140 sessions, while non-AI sessions total 501 credits across 437 sessions.

Looking beyond total credit volume, per-session averages provide a clearer comparison of educational weighting between AI and non-AI sessions.

This indicates that AI sessions are modestly more credit-dense than the programme average. While the absolute gap is not large, the consistent positive delta suggests AI content is not treated as peripheral innovation material. Instead, it carries slightly higher formal educational weighting per session, reinforcing its positioning within structured, accredited professional development.

Importantly, the difference is incremental rather than disproportionate. This balance implies integration rather than exceptionalism: AI is being embedded into mainstream educational tracks with marginally elevated depth, rather than isolated as a niche or promotional theme.

Overall direction

The ECR 2026 programme reflects a field in transition. AI is visible across research presentation, clinical subspecialties, operational governance themes, and formal education tracks. Clinical expansion and operational maturity are progressing in parallel rather than in sequence.

Taken together, the programme structure suggests that AI is no longer framed primarily as technological novelty. It is increasingly treated as an embedded component of contemporary radiology practice, with the most active discussions shifting toward integration, accountability, and real-world workflow impact.

For the exhibitor-side mirror of this integration trend, see ECR 2026 AI Industry Landscape.

Follow our ECR 2026 coverage

We will continue tracking high-density AI sessions, clinical deployment discussions, and real-world integration signals throughout ECR 2026. Subscribe to RadAI Slice for curated analysis, session highlights, and practical takeaways beyond headline announcements.

AI in radiology is moving fast. We focus on what is evidence-backed, operationally realistic, and clinically relevant.