ECR 2026 Launch Landscape: AI Moves Into Radiology Workflow

Good morning, there. At ECR 2026, 41 of 66 announced products involved AI, with workflow software and reconstruction tools dominating launches.

What stood out to me is how AI is no longer appearing mainly as standalone diagnostic tools. Instead, vendors are embedding AI directly into workflow platforms, reconstruction pipelines, and imaging systems. This signals a shift from experimental AI toward deployment-ready infrastructure, where AI quietly improves efficiency and image quality across the radiology workflow rather than acting as a separate product.

Here's what you need to know about Radiology AI last week:

• ECR 2026 Launch Landscape Shows AI Moving Into Workflow and Deployment

• FDA Updates AI Medical Device List Through 2025

• Deep Learning Accelerates and Improves Brain Metastasis MRI Diagnosis

• Mammography AI Quantifies Cardiovascular Risk in Women

• Plus: 4 newly released datasets, 6 FDA approved devices & 3 new papers.

📊 ECR 2026 Launch Landscape Shows AI Moving Into Workflow and Deployment

RadAISlice: Following our previous analyses of AI sessions and AI exhibitors at ECR 2026, we examined vendor announcements to understand how AI products launched around the conference reflect broader industry trends.

The details:

• 66 product launches were tracked across press releases and vendor announcements tied to ECR 2026.

• 41 launches were AI-related, representing about 62% of all product announcements.

• Software and platform products dominated with 56% of launches, compared with 27% for scanner hardware.

• Workflow AI and image reconstruction together accounted for about 70% of AI-related launches.

• MRI and X-ray were the most common technology areas among AI launches.

• Imaging OEMs remained the main drivers of announcements, responsible for roughly 60% of product launches.

Key takeaway: After examining AI sessions and exhibitors in earlier issues, the ECR 2026 launch landscape confirms the same shift: AI is increasingly embedded into workflow platforms and imaging systems rather than appearing as standalone diagnostic tools.

📊 FDA Updates AI Medical Device List Through 2025

RadAISlice: Radiology continues to dominate FDA AI clearances as the agency’s updated dataset now tracks authorizations through December 2025.

The details:

• FDA has authorized 1,451 AI-enabled medical devices since tracking began in 1995.

• Radiology accounts for 1,104 approvals, representing about 76% of all AI medical device authorizations.

• Radiology captured 75% of approvals in 2025, compared with 73% in 2024 and 80% in 2023.

• Major imaging vendors lead the regulatory landscape, with GE HealthCare topping the list at 120 radiology AI authorizations.

Key takeaway: Radiology continues to dominate FDA-cleared AI devices, with large imaging vendors accumulating approvals and consolidation shaping the competitive landscape of clinical AI. At RadAI Slice, we maintain a structured database of FDA-cleared imaging AI tools and are currently analyzing the data to understand how the regulatory landscape is evolving across vendors, modalities, and clinical tasks.

🧠 Deep Learning Accelerates and Improves Brain Metastasis MRI Diagnosis

RadAISlice: AI-enhanced MRI not only improves detection, but slashes interpretation time by 31%.

The details:

• Trained/validated on over 1,300 brain cancer MRI cases.

• AI read time: 100s/case vs. 144s for radiologist alone.

• Lesion sensitivity rose from 68% to 92%, AUROC from 0.84 to 0.95.

• Junior radiologists’ sensitivity increased by nearly 25%.

Key takeaway: Integrating this model can elevate diagnostic accuracy for both junior and senior readers and meaningfully improve efficiency in brain metastasis MRI reporting.

🩺 Mammography AI Quantifies Cardiovascular Risk in Women

RadAISlice: AI-derived BAC on routine mammograms predicts future major cardiac events in women.

The details:

• Data from 123,762 women across Emory and Mayo, spanning multiple races.

• Severe BAC linked to up to 3.3x increased risk of major cardiovascular events.

• Dose-response: +2-3% event risk for every 1 mm² BAC increase (p<0.001).

• BAC prognosticates beyond the PREVENT calculator in both cohorts.

Key takeaway: This research enables radiologists to provide actionable cardiovascular risk using standard mammograms, positioning breast screening as a dual-purpose preventive tool without added imaging or dose.

PETWB-REP (2026-03-09)

Modality: PET/CT | Focus: Whole-body, Multi-cancer | Task: Segmentation, Report generation

• Size: 490 scans, 490 patients

• Annotations: Paired radiology reports (bilingual, de-identified); structured clinical metadata

• Institutions: Shanghai Universal Medical Imaging Diagnostic Center, Fudan University et al.

• Availability:

  Public (Zenodo link)

• Highlight: Multi-cancer whole-body PET/CT with paired bilingual radiology reports; BIDS-formatted and fully de-identified.

Malmö Breast ImaginG (M-BIG) (2026-03)

Modality: DBT, DM | Focus: breast | Task: segmentation, detection

• Size: 104,791 women, ~500,000 exams (DBT+DM), 158 cancer cases

• Annotations: Detailed lesion outlines, morphological type (mass, calcification, etc.), BI-RADS

• Institutions: Lund University, Skåne University Hospital, et al.

• Availability:

  Public (tool): Mathworks File Exchange link; dataset availability unspecified

• Highlight: Unique, detailed lesion outlines for DBT and DM; supports combined radiomics, segmentation, and paired analysis.

SMC-LUD (2026-03-03)

Modality: US | Focus: Liver | Task: Classification

• Size: 5,385 images, 1,021 patients

• Annotations: Class labels (HCC or hemangioma), HCC confirmed by pathology, hemangioma radiologically diagnosed; metadata includes patient info and tumor stage

• Institutions: Samsung Medical Center, Sungkyunkwan University, et al.

• Availability:

  Public (Figshare link)

• Highlight: Largest public B-mode liver US dataset with HCC cases pathologically confirmed; supports robust AI liver lesion classification.

MRKR (Emory Knee Radiograph Dataset) (2024)

Modality: X-ray | Focus: Knee, Musculoskeletal | Task: Classification, Prognosis

• Size: 503,261 radiographs from 83,011 patients

• Annotations: Pain scores, ICD and CPT codes, laterality, view, weight-bearing, hardware, Kellgren-Lawrence grade (some AI-inferred)

• Institutions: Emory University, University of Florida, et al.

• Availability:

  Request-only (contact author, no public link provided)

• Highlight: Largest, most diverse knee X-ray dataset with pain scores and detailed clinical outcomes; 40% African American patients.

🏛️ FDA Clearances

• K253535 - Ligence Heart automates echocardiography image analysis, helping clinicians with cardiac function assessment.

• K253578 - Aidoc’s BriefCase-Triage: CARE Multi-Triage CT flags pneumothorax, pericardial effusion, aneurysm, and shoulder fracture on CT.

• K260217 - Exo Imaging’s AI Platform 2.2 (AIP002) enhances radiological image workflow efficiency and accuracy across modalities.

• K253244 - AirRay-mini series delivers mobile, portable X-ray imaging for clinical use in radiology.

• K253486 - SKIA-Head (SKIA-ST00) supports radiologists with enhanced processing/analysis for cranial images.

• K253714 - Philips IntraSight Plus supports cardiovascular clinicians by enabling advanced ultrasound-based imaging during interventions.

• Explore last week's 9 radiology AI FDA approvals.

📄 Fresh Papers

• doi:10.1093/eurheartj/ehag128 - BAC quantified by AI on >120,000 mammograms predicts cardiovascular events with hazard ratios 1.3–3.3, outperforming risk calculators.

• doi:10.1038/s41597-026-07058-w - A 490-patient, multi-cancer public PET/CT dataset supports radiomics, report generation, and multimodal learning research.

• doi:10.1007/s00330-026-12395-w - The ESUR Prostate MRI Working Group proposes a three-step global roadmap for standardized prostate MRI quality and innovation.

• Browse 155 new radiology AI studies from last week.

📰 Everything else in Radiology AI last week

• AI using routine mammograms quantifies BAC and stratifies CVD risk across two US health systems.

• Google’s mammography AI matches or exceeds radiologists as a second reader, increasing efficiency in NHS breast cancer screening.

• ECRI ranks AI diagnostic challenges in imaging as leading the 2026 patient safety agenda, warning about missed critical conditions.

• Physician AI adoption doubles to 81%, but most remain cautious on patients using AI to interpret radiology reports.

• Read 21 fresh radiology AI news articles for last week.

That's it for today!

👋 Quick favor: drag this into your Primary tab so you don’t miss next week. Or just hit Reply with one thought.

See you next week.

P.S. We keep building free tools to accelerate your radiology work. What's the most time-consuming pain point in your day that we should help speed up? Reply and share your take so we keep building around you.