AI Medical Imaging Companies: A Structured Landscape for Healthcare Decision-Makers

Executive Summary: The AI Medical Imaging Company Market in 2026

The AI medical imaging company market has crossed a critical inflection point. What began as a collection of single-algorithm startups competing on sensitivity and specificity for a narrow finding — a pulmonary embolism here, a pneumothorax there — has consolidated into a structured ecosystem with distinct tiers, enterprise platform strategies, and a growing divide between capital-rich leaders and specialized niche players.

As of mid-2026, the U.S. Food and Drug Administration has authorized 1,104 radiology AI devices, representing 76% of all AI-enabled medical device authorizations since 1995. A small group of well-capitalized companies — Aidoc, Viz.ai, Lunit, and Annalise.ai — hold the majority of those clearances, have raised hundreds of millions in venture funding, and are deployed across thousands of hospitals. At the same time, a second wave of workflow-first startups is redefining competition by embedding AI into PACS, reporting automation, and full radiology service models rather than selling algorithms as standalone products.

This article provides a structured, evidence-based reference for hospital administrators, radiology directors, health IT procurement teams, and investors who need to understand the current landscape — not as a list of vendors, but as a map of market structure, regulatory standing, clinical focus areas, and the practical criteria that differentiate one category of company from another.

Market Overview: Size, Growth, and Regulatory Context

The global AI in medical imaging market was valued at approximately $1.8 billion in 2025 and is projected to reach $20.2 billion by 2033, representing a compound annual growth rate (CAGR) of roughly 35%. These figures come from market research firms that define the market as including AI software for image acquisition, reconstruction, interpretation, triage, and workflow orchestration across radiology subspecialties.

The regulatory environment provides the most concrete measure of market maturity. According to the FDA's updated list through the end of 2025, 1,104 of the 1,451 total AI-enabled medical devices authorized since 1995 are radiology devices — a 76% share. In Q4 2025 alone, the FDA cleared 72 AI-enabled medical devices, of which 55 (76%) were radiology. For all of 2025, radiology secured 75% of authorizations, compared to 73% in 2024 and 80% in 2023.

Tier 1: Market Leaders — Aidoc, Viz.ai, Lunit, Annalise.ai

Four companies have emerged as the dominant pure-play AI medical imaging vendors by nearly any measure: FDA clearance count, venture capital raised, hospital deployment scale, and breadth of clinical indications covered.

Aidoc holds 31+ FDA clearances and has raised approximately $250 million. Its platform covers multi-condition detection across emergency CT workflows — including intracranial hemorrhage, pulmonary embolism, rib fractures, and cervical spine injuries — and is deployed in hundreds of hospitals globally. In 2025, Aidoc received FDA clearance for a rib-fracture AI that uses a foundation-model approach, signaling a shift from single-task algorithms to more flexible, multi-condition architectures.

Viz.ai is deployed in more than 1,700 hospitals across the U.S. and Europe, making it the most widely adopted AI imaging platform by site count. Its core focus is stroke care coordination: AI-powered detection of large vessel occlusion (LVO) on CT angiography, followed by automated alerts and communication workflows. Two studies presented at the 2025 International Stroke Conference reported that Viz LVO implementation reduced time from patient arrival to LVO diagnosis by 44% and reduced treatment time by an average of 31 minutes in a multicenter study of 474 patients.

Lunit has differentiated itself through OEM embedding: its breast cancer AI is integrated on GE HealthCare, Philips, and Fujifilm mammography systems, giving it distribution reach that pure-software vendors struggle to match. The company also offers chest X-ray and chest CT AI products and has raised substantial venture funding to support global expansion.

Annalise.ai takes a different approach: its chest X-ray AI can detect up to 124 findings on a single image, positioning it as a comprehensive screening tool rather than a focused triage algorithm. This breadth of detection creates a different value proposition for health systems that want one AI covering multiple chest pathologies rather than stacking multiple single-condition algorithms.

For a deeper dive on Aidoc's specific products, funding history, and deployment case studies, see our dedicated Aidoc: Radiology AI Company Profile.

Tier 2: Specialized Players — Gleamer, Subtle Medical, Rad AI, Qure.ai, Cleerly, HeartFlow

Below the Tier 1 leaders, a group of well-funded, FDA-cleared companies has carved out defensible positions in specific clinical niches or workflow segments. These companies may not match Aidoc or Viz.ai in breadth of clearances or deployment count, but they lead in their respective domains.

Subtle Medical focuses on MRI acceleration. Its AI solutions denoise and sharpen MRI scans, allowing facilities to conduct scans up to 80% faster. The company has 11 FDA clearances and is deployed on more than 1,300 scanners across 600+ sites, serving approximately 2.5 million patients per year. In June 2026, Subtle Medical secured $33 million in Series C funding, bringing total funding to $86 million. Customers include Mount Sinai, RadNet, and Radiology Partners. The company is also developing SubtleGAD, designed to enable up to 90% reduction in gadolinium contrast dose.

Rad AI addresses a different bottleneck: radiology reporting. Rather than competing on diagnostic accuracy for a specific finding, Rad AI's platform automates report generation, impression drafting, and follow-up recommendation tracking. This workflow-first approach targets radiologist burnout and report turnaround time — operational metrics that matter as much to department directors as sensitivity and specificity.

Qure.ai has built a strong presence in chest X-ray and CT interpretation, particularly for tuberculosis screening, lung nodule detection, and trauma assessment. The company is one of the four that captured 72% of the $615 million in 2022 startup funding, alongside Aidoc, Viz.ai, and Cleerly, reflecting investor confidence in its global market potential.

Cleerly and HeartFlow operate in the cardiac imaging subspecialty. Cleerly's AI analyzes coronary CT angiography to quantify plaque burden and characterize stenosis. HeartFlow's flagship product uses computational fluid dynamics applied to coronary CT to derive fractional flow reserve (FFR-CT), a non-invasive alternative to invasive catheterization. Both companies have FDA clearance and are increasingly referenced in cardiology guidelines.

Emerging Platform and Orchestration Players — deepc, Blackford, EnvoyAI

A distinct category has emerged in the last two to three years: AI orchestration platforms that do not develop their own algorithms but instead provide the infrastructure for hospitals to integrate, manage, and deploy multiple AI applications from different vendors through a single interface.

deepc offers an enterprise AI integration and multi-vendor orchestration platform that connects radiology departments to a marketplace of FDA-cleared algorithms. Instead of negotiating separate contracts, IT integrations, and support agreements with each AI vendor, a hospital can use deepc's platform to access algorithms from multiple companies through a single workflow layer. This model addresses a real operational pain point: as the number of AI algorithms in a department grows, managing them individually becomes unsustainable.

Blackford Analysis and EnvoyAI (a division of IBM Watson Health) operate similar marketplace models. These platforms are particularly relevant for health systems that want to pilot multiple AI tools without committing to a single vendor's ecosystem, or that need to swap algorithms as new evidence emerges without re-engineering their PACS integration.

• Single integration point for multiple AI vendors
• Centralized management of algorithm deployment, updates, and decommissioning
• Reduced IT burden for radiology departments scaling AI adoption
• Ability to compare algorithm performance on local data before committing

Digital Pathology AI: A Smaller but Fast-Growing Adjacent Market

While the primary focus of this article is medical imaging (radiology), the adjacent digital pathology AI market deserves mention as a fast-growing segment that follows a similar market structure. The AI in pathology market was valued at approximately $135 million in 2024 and is projected to reach $1.15 billion by 2033, growing at a CAGR of roughly 27%.

Key companies in this space include:

Digital pathology AI is smaller than radiology AI in both market size and FDA clearance count, but it follows a similar trajectory: early point solutions for specific cancer types are giving way to broader platforms that cover multiple stains, tissue types, and workflow stages.

Incumbent OEMs: GE HealthCare, Siemens Healthineers, Philips, Canon, United Imaging

No landscape of AI medical imaging companies is complete without the major imaging OEMs, which hold the largest number of radiology AI authorizations — though for reasons that require careful interpretation.

According to the FDA's authorization data, GE HealthCare leads with 120 radiology AI authorizations, followed by Siemens Healthineers (89), Philips (50), Canon (45), and United Imaging (38). These counts include both standalone software algorithms and imaging hardware with embedded AI capabilities — a distinction that matters for procurement decisions.

The OEM strategy has been to acquire AI startups and embed their algorithms directly into scanner workflows. GE HealthCare's acquisitions of Caption Health (cardiac ultrasound AI), MIM Software (therapeutic imaging AI), and icometrix (neuroimaging AI) illustrate this approach. For health systems, the trade-off is clear: OEM-embedded AI offers seamless integration and single-vendor accountability, but it may limit flexibility to choose best-in-class algorithms from multiple vendors.

A Practical Vendor Selection Framework by Organizational Priority

Not every health system needs the same AI vendor. The right choice depends on the organization's primary clinical and operational priorities. The following framework maps four common institutional priorities to the vendor categories most likely to meet them.

Additional criteria to consider during evaluation:

• FDA clearance count and type (510(k) vs. De Novo) — indicates regulatory maturity and breadth of intended use
• Evidence tier — peer-reviewed RCT, prospective study, retrospective study, or white paper only
• EHR/PACS integration — confirmed integrations with Epic, Oracle Health, or major PACS vendors
• Deployment scale — number of hospitals or scanners where the AI is actively used
• Company funding stage and financial stability — relevant for long-term procurement commitments

Key Trends Shaping the Market: Consolidation, Foundation Models, and Reimbursement

Three structural trends will define the AI medical imaging company market over the next three to five years.

Consolidation Through Acquisitions and Platform Plays

The market is consolidating. OEMs are acquiring AI startups to embed into their hardware. Tier 1 companies are expanding their algorithm portfolios through internal development and acquisition. AI orchestration platforms are positioning themselves as the neutral layer that prevents any single vendor from owning the radiology AI stack. The result is a market where standalone, single-algorithm startups face increasing pressure to either scale rapidly, find a niche defensible against platform competition, or be acquired.

Foundation Models Enable Multi-Condition Algorithms

Aidoc's 2025 FDA clearance for a rib-fracture AI built on a foundation-model approach marks a significant technical shift. Instead of training separate models for each finding, foundation models learn general representations of anatomy and pathology that can be adapted to multiple tasks with less labeled data. This approach could compress the time and cost of developing new algorithms and enable a single AI to detect dozens of conditions — challenging the current paradigm of stacking multiple single-condition algorithms.

Reimbursement Progress and Regulatory Modernization

Two regulatory developments are worth tracking closely. First, the Centers for Medicare & Medicaid Services (CMS) now reimburses AI coronary plaque analysis (as of 2025), providing a precedent for broader AI reimbursement that could unlock adoption beyond the current grant-funded and ROI-justified purchasing models.

Second, the FDA has formalized Predetermined Change Control Plans (PCCPs), which allow AI developers to pre-specify how their models will be updated over time — meaning routine modifications no longer trigger new 510(k) submissions. This regulatory modernization is critical for AI products that improve continuously, and it reduces the regulatory risk for hospitals adopting AI from companies with active PCCPs.

• CMS reimbursement for AI coronary plaque analysis (2025) — precedent for broader AI coverage
• FDA Predetermined Change Control Plans (PCCPs) — enable continuous model updates without new submissions
• Foundation model approach — reduces development cost for multi-condition algorithms
• OEM acquisition spree — GE, Siemens, Philips, Canon actively acquiring AI startups
• Second-wave workflow-first startups — embedding AI into PACS, reporting, and radiology-as-a-service models

For a deeper analysis of the clinical evidence supporting AI imaging tools — including study methodology, dataset diversity, and known limitations — see our companion article: AI Medical Image Analysis: A Critical Appraisal of the Evidence Base.