HOPPR, a medical imaging AI company, has opened applications for the next cohort of its Catalyst Program at the SIIM26 Annual Meeting in Chicago. The programme provides clinician scientists and researchers with access to foundation models, datasets, fine-tuning tools and machine learning support via the HOPPR AI Foundry platform. The first cohort includes projects from MAI Lab Lagos developing mammography AI models for breast cancer detection in Africa, The Catholic University of Korea adapting chest X-ray models to identify musculoskeletal abnormalities, and the University of Illinois Cancer Center exploring lung and breast cancer risk prediction. Founded in 2019, HOPPR aims to lower technical barriers in healthcare AI development. The company is showcasing interactive demonstrations at booth #509 during SIIM26 and will host an informational webinar in July.

Platform control becomes owned infrastructure; screening AI shows real labor proof - march 20, 2026. GE's Intelerad close turns enterprise imaging strategy into balance-sheet reality, while new screening evidence and cyber/helium disruptions sharpen what buyers will actually fund One big thing. Enterprise imaging crossed from conference narrative into owned infrastructure this week: GE's Intelerad close made platform control tangible, new breast-screening AI data made labor relief more credible, and cyber/helium shocks kept resilience at the center of buying decisions. Listen to this week's Marketstrat Pulse Insight: Key takeaways. * Enterprise imaging control moved from messaging to ownership. GE HealthCare's Intelerad close is the week's most material signal because it puts a cloud-first enterprise imaging layer inside an OEM stack, raising bundling leverage, recurring-software exposure, and switching-cost risk for providers and independent AI vendors. * Breast AI finally delivered a prospective labor signal that matters. The Nature Medicine trial showed materially lower radiologist workload with higher cancer detection, but higher recall means the economics are still pathway-dependent. This is evidence for workflow redesign, not effortless margin expansion. * Pediatric imaging remains underbuilt as a commercial and regulatory category. The JAMA labeling analysis and OXOS's pediatric clearance expansion point in the same direction: low-friction workflow tools can move faster than pediatric-specific AI categories that require heavier evidence, labeling, and liability work. * The model layer is commoditizing faster than the workflow layer. HOPPR's NVIDIA integration, Circle's vascular CT expansion, and GE's Springbok collaboration all reinforce the same commercial truth: distribution and installed workflow matter more than open-model novelty by itself. * Reimbursement was quiet, not solved. In the accessible strict-window source set, no material new imaging-specific CMS/MAC LCD/NCD/CPT action surfaced. That keeps enterprise contracting, throughput math, and operational ROI as the primary monetization path for imaging AI. * MIS and medtech resilience still shape imaging budgets. Imperative Care's financing and Stryker's cyber disruption both reinforce that procedural ecosystems and operating resilience remain live competitors for the same hospital capital envelope. Innovation hook. Breast screening AI finally produced prospective labor proof - but not frictionless economics The week's highest-signal evidence item was not another accuracy claim. It was a prospective workflow result. In the Nature Medicine study, the AI strategy cut radiologist reading volume sharply while increasing screen-detected cancers but recall also moved higher. That makes the commercial lesson unusually clear: screening AI can create real labor relief, yet the economic case still depends on what happens downstream. A buyer cannot underwrite the value of fewer reads in isolation if workup volume, callbacks, or pathway complexity rise at the same time. In practical terms, this is evidence for operating-model redesign, not plug-and-play margin expansion. Data basis: Study-reported radiologist readings, cancer detection, and recall indexed against standard workflow. Prospective breast-AI evidence now points to a real labor-capacity lever. The catch is that recall still matters commercially, so the value case is workflow redesign, not just "better detection." Market lens - North America ultrasound market. Care model divergence: Same region, different delivery logic North America should be read as a mature installed-base, workflow- and labor-driven ultrasound region, not as an access-creation market. The region is still overwhelmingly U.S.-driven in scale. By 2035E, the U.S. represents about 91% of North America systems revenue, 91% of ecosystem revenue, and roughly 91% of unit shipments. Canada is much smaller in scale, but it matters strategically because it changes the regional care-model mix and highlights that North America is not one single commercial model. The key takeaway is that the two countries are directionally aligned on technology, but not identical in delivery logic. In both markets, compact / portable becomes the largest revenue engine by 2035E and handheld / POCUS becomes the fastest shipment-growth engine. Where they diverge is in site of care, procurement logic, and monetization intensity. Signal Pulse heatmap - mar 14 - 20, 2026. Event-level The heatmap shows a week where platform control and clinical evidence shared the top tier. GE's Intelerad close and the new breast-AI trial both scored at the structural end of the range because they affect how buyers think about ownership and labor. Mid-tier signals clustered around financing, governance, quantification, and operational reality checks. The pattern matters: the market did not reward raw algorithm novelty this week. It rewarded events that either changed control of the deployment surface or improved clarity on whether AI actually alters operating models. That is a more mature signal pattern than the one-off clearance-heavy weeks seen earlier in the year. The strongest signals this week were ownership and operating-model signals, not feature releases. Platform control and evidence quality dominated the score distribution. Quick-glance table. This week's Pulse research note also covers: GE HealthCare's Intelerad close and the shift from enterprise imaging strategy to owned infrastructure; breast screening AI evidence and the real economics of labor relief versus downstream recall burden; FDA and governance signals including pediatric imaging gaps, portable X-ray expansion, and MR-guided breast biopsy workflow advances; platform-control moves across imaging IT, open-model tooling, and vascular CT analysis; MRI operational risk and medtech resilience, including helium sensitivity and cyber-related disruption; minimally invasive surgery funding and procedural ecosystem competition; and provider, OEM, payer, and AI-vendor read-through across workflow, distribution, and capital allocation. About Marketstrat Marketstrat(R) is a market intelligence and GTM enablement firm committed to empowering clients in data-driven industries. Under the Markintel(TM) brand, it delivers robust market intelligence, while GrowthEngine solutions offer specialized GTM advisory and app-based tools - together fueling growth, innovation, and competitive advantage. For more information, visit www.marketstrat.com. Marketstrat(R) is a registered trademark and Markintel(TM) is a pending trademark of Marketstrat.

HOPPR has integrated NVIDIA's NV-Reason and NV-Generate open models into its AI Foundry platform for medical imaging development, announced at NVIDIA GTC 2026. The HIPAA-compliant platform combines accelerated computing, curated datasets and foundation models for building imaging AI applications. NV-Reason provides multimodal reasoning for chest X-ray interpretation, generating structured analytical steps alongside outputs for greater transparency. NV-Generate creates synthetic DICOM imaging datasets to support model training and validation where real-world data is limited. Built on NVIDIA A100 and H100 GPUs, the Foundry enables developers to train and fine-tune medical imaging models using optimised infrastructure. HOPPR's Forward Deployed Services offers expert support for model customisation, combining machine learning engineers, data scientists and clinical experts to refine imaging applications.