Radiology & Imaging AI

Philips Lumify is indexed as a product rather than as Royal Philips, whose portfolio is far too broad to grade on these axes; the assessment covers Lumify and the software that runs on it, and Philips' own pages are its estate. Lumify is a cabled transducer that plugs into a smart device the customer already owns, over USB on Android or Lightning on Apple hardware, with the app downloaded from the ordinary app stores. There is no proprietary tablet, no wireless pairing and no battery in the probe, which is why scanning runs to roughly 7.5 hours on a suitable Android device. Three broadband transducers cover the range (S4-1 phased array, C5-2 curved array and L12-4 linear), and fourteen presets include abdominal, cardiac, lung, FAST, musculoskeletal, obstetric and ocular. Pulsed wave Doppler arrived in 2022, and Tissue Harmonic Imaging, SonoCT and xRes carry over from the cart line. Lumify was introduced in 2015 (K152899); the US product page now sells it outright with no subscription and all software included. Two artificial intelligence features ship, both narrow and both Android only. Auto B-line Quantification counts B lines frame by frame and detects merged B lines inside a guided 12 point lung protocol; its expanded version was cleared in May 2023 (K223771) as a locked machine learning classifier, developed under a partnership with the US Biomedical Advanced Research and Development Authority, and tested on 416 video loops from 157 adults with shortness of breath (merged B line sensitivity 0.83, specificity 0.92). Auto EF calculates left ventricular ejection fraction from a single apical four chamber view, and its clearance record identifies the underlying model as the LVivo EF algorithm developed by DiA. Tele ultrasound runs through Reacts, a platform from Innovative Imaging Technologies, so a remote colleague can see the live scan and the device camera at once while the scanning clinician controls what is shared. The device is used in austere settings, including a Philips Foundation program in Kenya training village midwives to perform antenatal screening with remote radiologist support. Three things a buyer should weigh. Both AI features are Android only, so a buyer standardized on Apple hardware gets the scanner without the models, and the product page does not mention the AI features at all. No price is published for any transducer. And Philips publishes a mature product security disclosure program and product level data security documentation, but no audited security certification and no business associate position for the product's cloud and tele ultrasound functions.

Vscan Air is indexed as a product rather than as GE HealthCare, whose portfolio is far too broad to grade on these axes; everything here concerns Vscan Air and the software sold on it, and GE HealthCare's pages are its estate. The device is a wireless dual probe that pairs to a phone or tablet and fits in a coat pocket. Two configurations exist: the CL couples a convex and a linear transducer for obstetric, gynecologic, abdominal and musculoskeletal imaging, and the SL couples a sector phased array and a linear transducer for cardiac, lung, vascular access and musculoskeletal work. B mode, M mode, color Doppler and pulsed wave Doppler ship with the probe and work without a subscription. Vscan Air launched in 2021 as the current generation of a handheld line GE began with Vscan in 2010, with probe clearances including K202035, K231301 (2023) and K250087 (2025). The AI arrived by acquisition. GE HealthCare bought Caption Health in 2023 and brought Caption AI to Vscan Air SL in April 2024 as a purchased option. Caption Guidance gives real time prescriptive instruction toward standard echocardiographic views and captures automatically once quality is reached; it was authorized through De Novo in February 2020 (DEN190040) as the first AI permitted to guide the acquisition of medical images. Caption Interpretation AutoEF calculates left ventricular ejection fraction from one to three views without manual border tracing (K210747, January 2022; root mean square deviation 7.21 ejection fraction points against a biplane reference in 186 studies). The pivotal evidence is public: a registered, prospective trial in JAMA Cardiology (2021) in which eight nurses with no prior echocardiography experience scanned 240 patients across two hospitals, with five blinded echocardiographers judging images of diagnostic quality for left ventricular size and function in 98.8 percent of patients. Several authors were employees of the company whose product was tested, and the trial ran on predecessor hardware. Three things a buyer should weigh. The AI is an option rather than a property of the device: cardiac only, sold separately, on the SL and not the CL, and not available in every country. Cloud storage and image sharing sit in a separate tier. And the product terms let GE HealthCare use non personal data related to its products for training and product improvement and own the results, while undertaking not to sell the data or identify the customer; what counts as non personal data is not defined.

EchoNous makes Kosmos, a handheld ultrasound system positioned against cart based systems rather than pocket scanners. Its case rests on hardware: proprietary piezoelectric transducers, proprietary application specific circuits, and continuous wave Doppler, which the company states no other handheld offers and which allows valve gradients and pressure calculations at the bedside. Kosmos runs on iOS, Android and a dedicated tablet, and is sold in more than 70 countries with a five year warranty. The artificial intelligence is a set of workflow models the company describes as proprietary deep learning that helps novice users learn ultrasound quicker. AI FAST labels anatomy and identifies views in real time, an assisted ejection fraction workflow produces an automated left ventricular measurement, Auto Preset switches the exam preset as the operator moves between abdomen, lung and heart, and Auto Doppler positions the sample gate on the valve. A February 2026 release was announced as adding cleared obstetric biometry and billable vascular workflows, and a June 2026 release added Vascular Auto Flow for dialysis fistula monitoring, tricuspid valve Doppler calculations and DICOMweb. Two partner services are named on the platform: Us2.ai for automated echocardiography reports and 19Labs for tele ultrasound streaming. EchoNous holds five 510(k)s as applicant, from the Uscan bladder scanner (K181574, 2018) through Kosmos clearances in 2020, 2021, 2024 and 2025. The 2021 summary states that the AI assisted ejection fraction workflow and the Trio tool were then not yet cleared and were offered under FDA's COVID-19 enforcement policy for imaging systems; which features the 2024 and 2025 clearances cover is what a buyer should confirm. The evidence for the automated ejection fraction is largely independent: a prospective multicenter Japanese study with separate core laboratories, a diagnostic accuracy study in npj Digital Medicine, an Italian cohort measured against cardiac magnetic resonance, and a cardio oncology cohort scanned by staff without ultrasound expertise, with the Japanese study reporting that the system tended to underestimate ventricular volumes. The company also publishes its own prospective 78 patient core laboratory study. Data handling is architectural: the products privacy policy states that patient data is accessed only locally on the device and that EchoNous does not access, collect or disclose it, and a Chief Privacy Officer is named. Scanning works offline. The company was founded in Redmond, Washington, by Kevin Goodwin, who earlier founded SonoSite, and Dr Niko Pagoulatos, and began as Signostics, a bladder scanner business. Its own account dates the founding to 2016, while its corporate registration and site footer point to 2015. It is privately held, backed principally by KKR with later financing from Kennedy Lewis, and reports roughly $194 million raised. Two things a buyer should weigh. The home page calls the product HIPAA and SOC2 compliant, while no SOC 2 report, security page or vulnerability disclosure policy is published for a networked device that takes over the air updates. And pricing is by quote, with one configuration stated at under $20,000 and no statement of whether a recurring fee applies.

Exo, pronounced echo, makes the Exo Iris handheld ultrasound probe and describes itself as a medical imaging software and devices company. Its transducer is a piezoelectric micromachined design built on silicon, blending the imaging behavior of piezoelectric crystals with the cost structure of a chip. The Exo Iris was cleared in 2021 (K211527), with imaging modes and indications added in 2022 (K222198), and launched commercially in September 2023 at a starting price of $3,500, sold through the company's online store and its sales team. The artificial intelligence sits on the probe as suites: Cardiac AI (automated ejection fraction, chamber size, strain and wall thickness), Lung AI (B lines, effusions, consolidations and lung pathology in real time), Fluid Management AI (IVC collapsibility, bladder volume and vascular access) and SweepAI, which gives instant feedback as the user sweeps. The company states these are FDA cleared, reported 14 cleared AI indicators by June 2025, and describes its June 2025 clearance for detecting pleural effusion and consolidation or atelectasis as the first for point of care ultrasound AI, stating that it showed superiority to clinicians alone. Exo Works is the workflow half, launched in 2022: vendor neutral software for documentation, billing, quality assurance and education in point of care ultrasound programs, described as integrating with nearly any point of care ultrasound device and with hospital record and archive systems. Wellstar MCG Health, formerly Augusta University Health, implemented it across its health system. The company announced in 2023 that Exo Works earned HITRUST r2 certification and a SOC 2 Type 2 report, and it links a Security Trust Center from its site. Exo reports about $428 million raised, most recently in May 2025. It is led by co founder and chief executive Sandeep Akkaraju and headquartered at 4201 Burton Drive, Santa Clara, California. Its stated users include physicians, nurses and emergency medical technicians in emergency, acute, outpatient, home and rural settings. Two things a buyer should weigh. The company markets a detection feature as outperforming clinicians, and the study's design and figures are not on its home page. And the recurring terms are unpublished: a device price is public, while whether the AI suites need a subscription, and how Exo Works is priced, are not.

Clarius Mobile Health makes wireless handheld ultrasound scanners and the AI that runs on them. Seven scanners cover curved array abdominal and obstetric work, phased array cardiac and eFAST, linear vascular, musculoskeletal and nerve imaging, ultra high frequency aesthetics, and endocavity gynecology and urology, each connecting directly to an iOS or Android device; there is also a veterinary line. Clarius Intelligence adds models that each carry their own FDA clearance: MSK AI (2023), Bladder AI (2023), OB AI for fetal biometry (2024), Prostate AI and Median Nerve AI (2025) and Ejection Fraction AI (2026), plus automatic preset selection, voice control and T-Mode anatomy visualization for education and procedural planning. Scanners are sold on the company's online store from $4,490 to $6,090 including a year of membership, and the scanner works for core imaging without a membership, which adds cloud storage, AI models and team tools. Studies can stay on the device or go to Clarius Cloud, stored in Canada or a region the customer selects. The company reports more than 7 million scans in over 70 countries, among emergency, primary care, obstetric, musculoskeletal, urology, aesthetic and education users. Headquartered in Vancouver, British Columbia, and led by Ohad Arazi.

Butterfly Network makes handheld ultrasound probes that image with a semiconductor chip in place of piezoelectric crystals, so a single probe connected to a phone or tablet covers the whole body where conventional systems use several transducers. The current probe is the Butterfly iQ3 (K232808, January 2024), and the iQ+ is still sold, listed from $2,699 on the company's site. Because imaging runs in software, new capability reaches probes already in use through cloud delivered updates. The artificial intelligence sits in three layers. The company's own models assist acquisition and measurement, and two carry their own clearances: the Auto 3D Bladder Volume Tool (K200980) and the Butterfly Gestational Age Tool (K252148, March 2026), a deep learning model that estimates gestational age from blind sweeps guided by on screen animations, without the operator acquiring standard views. The company describes it as the first cleared blind sweep tool; its clearance study was a prospective evaluation of 111 pregnancies at four US sites, and it is cleared as adjunctive information for trained professionals, not for prenatal management or delivery planning. Compass AI is enterprise workflow software for point of care ultrasound programs, and Butterfly Garden is a partner marketplace through which other developers deliver their own applications on Butterfly hardware. ScanLab is an AI guided training application, and Butterfly Embedded licenses the chip technology to other manufacturers. Butterfly is publicly listed and headquartered in Burlington, Massachusetts; it reported first quarter 2026 revenue of $26.5 million. Buyers span hospitals and health systems, medical schools running one probe per student programs such as Indiana University School of Medicine, individual clinicians and small practices buying through the online store, veterinary practices, global health programs and home and community care. The University of Rochester reported a 116 percent increase in point of care ultrasound revenue after a systemwide deployment. Two things a buyer should weigh. Studies are stored in Butterfly's cloud: a SOC 2 Type 2 report is named on the security page, and no hosting region or business associate position is published. And the gestational age tool is cleared for trained professionals as adjunctive information, while it is marketed for settings where no sonographer is available.

contextflow reads chest CT and shows the radiologist comparable cases alongside its findings. ADVANCE Chest CT provides computer aided detection support for suspected lung cancer, interstitial lung disease and COPD: nodule detection, measurement and tracking over time, quantification of seven lung tissue findings, analysis of 19 image patterns, incidental coronary calcium, and a 3D image search that retrieves similar cases from a knowledge base of more than 8,000 anonymized lung CTs from partner hospitals, with links to differential diagnosis literature. Results appear in the radiologist's own viewer, and the software runs on a virtual machine inside the hospital's network. The company's suite also resells RevealAI-Lung (RevealDx) and CINA-iPE (Avicenna) and a lung cancer screening workflow built with Mint Medical for Germany's national program. A spinout of the Medical University of Vienna and the KHRESMOI research project, founded in 2016 in Vienna, it is CE marked under the EU medical device regulation and holds ISO 13485 certification; its nodule detection feature is not yet cleared in the US. A European Radiology study (2022) with the Medical University of Vienna reported 31 percent shorter report reading time. Users include Erasmus MC, Freiburg, Innsbruck, Leiden and Mainz. In June 2026 4DMedical Limited, an Australian listed respiratory imaging company, agreed to acquire contextflow for more than 11.4 million euros; the contextflow brand and site continue.

Belgian company, founded in 2011 in Leuven and now part of GE HealthCare, whose icobrain software measures the brain on MRI and CT. It segments and quantifies brain structures and lesions and reports volumes and change over time against an age and sex matched reference population. Modules are sold by condition: multiple sclerosis, dementia and Alzheimer's, amyloid related imaging abnormalities (icobrain aria), stroke, traumatic brain injury, epilepsy, Parkinson's disease and brain tumors. icobridge is a gateway in the hospital that pseudonymizes and encrypts scans before cloud analysis and returns DICOM, PDF and HL7 reports and prepopulated templates to PACS, RIS or the record system. icompanion is an app for people with multiple sclerosis. FDA lists six icometrix clearances, from icobrain in 2016 to icobrain aria in November 2024, and the software is CE marked. icometrix names ISO 27001 and ISO 13485 certificates, security questionnaires, and business associate and data processing agreement templates, all on request, and holds the ACR Transparent AI badge. It says it is used in more than 400 hospitals in over 30 countries, with offices in Leuven and Boston. Two Category III CPT codes apply in the US. No price is published.

Optellum, based in Oxford with a US office in Houston, sells the Virtual Nodule Clinic, which identifies patients with pulmonary nodules, triages and tracks them, and scores malignancy risk with its Lung Cancer Prediction AI, a convolutional neural network applied to the CT scan. The FDA cleared the Virtual Nodule Clinic in March 2021 (K202300) for incidentally detected solid and semi solid nodules of 5 to 30 mm in patients aged 35 and over, and the company reports UKCA and CE marking and Australian approval in February 2026. It describes its prediction AI as the only FDA cleared, reimbursable radiomics AI solution. Two temporary procedure codes for quantitative CT tissue characterization have applied since July 2022, and the company publishes how hospitals are paid under them; its own price is not published. Published evidence includes external validations in Thorax and the American Journal of Respiratory and Critical Care Medicine (2020), a real world impact study (BMJ Open Respiratory Research, 2025) and a US payer cost effectiveness analysis (2026). Named users include the University of Pennsylvania Health System, Vanderbilt University Medical Center, Oxford University Hospitals and Tanner Health. The company is building LungOS on a thorax CT foundation model announced in 2025.

Oxford spinout that sells AI imaging software for stroke and lung disease. Brainomix 360 Stroke reads non contrast CT, CT angiography, CT perfusion and MRI: it scores early ischemic change (ASPECTS), estimates core volume, and flags suspected large vessel occlusion and hemorrhage for the stroke team by PACS, email, web and mobile app. Brainomix 360 e-Lung quantifies features on chest CT for interstitial lung disease, and Vantage is a network analytics dashboard. Brainomix holds 13 FDA 510(k) clearances (2020 to 2026) and CE marking as Class IIb under the EU MDR, and says it runs in more than 400 hospitals in over 20 countries. Its own FDA summary for Triage Stroke publishes error rates and results by age, sex, slice thickness and scanner maker. Large vessel occlusion sensitivity on non contrast CT was 69.6 percent, lower in older patients. The company reports a Lancet Digital Health study of 452,000 patients at 107 hospitals associating its stroke software with a doubling of thrombectomy rates. It is delivered as a managed appliance on hospital or cloud servers, is ISO 27001 certified, and licenses stroke modules individually or as a package. It also sells imaging biomarkers as clinical trial endpoints to pharmaceutical partners. Offices in Oxford, Ireland and the United States. No price is published.

Genius AI Detection is the breast imaging AI software line of Hologic, Inc. of Marlborough, Massachusetts; this record covers the software a breast imaging buyer licenses, not Hologic's wider diagnostics, surgical and skeletal businesses. The line comprises Genius AI Detection, a deep learning detector for digital breast tomosynthesis (K201019, November 2020); Genius AI Detection 2.0 (K221449, October 2022, with later clearances K230096 and K243341), which adds a convolutional neural network for calcification clusters and lesion correlation across views; Genius AI Detection PRO, a 2D and tomosynthesis platform with prior exam aggregation, automated pre reporting, image quality checks and a case score from 1 to 10; Quantra breast density assessment; and 3DQuorum, which generates 6 mm SmartSlices. The predecessor, ImageChecker CAD, is conventional 2D computer aided detection. The software reads images from Hologic's own Dimensions and 3Dimensions systems, returns results to review workstations and archives as DICOM objects, and in the PRO version uses cloud processing for prior exams. The tomosynthesis platform is approved under premarket approval P080003 (2011), and Hologic's gantries are also the acquisition hardware many competing breast AI algorithms read. Hologic reports a ruling in April 2026 by the Unified Patent Court in Germany that a competitor infringed a Hologic mammography patent. Hologic was taken private on 7 April 2026 by funds managed by Blackstone and TPG, with minority investments from ADIA and GIC, for up to $79 per share including a contingent value right tied to Breast Health revenue in fiscal 2026 and 2027, and has ended public financial reporting.

DeepHealth is RadNet's digital health business and the entity that contracts RadNet's clinical AI and imaging informatics software to third party providers. It is a wholly owned subsidiary of RadNet, Inc. (NASDAQ: RDNT), headquartered in Somerville, Massachusetts, and the brand was launched in November 2023 to consolidate acquisitions: eRAD (radiology information and archiving systems), the original DeepHealth breast AI company, Aidence (lung), Kheiron (breast), Quantib (prostate and brain), See-Mode (thyroid and vascular ultrasound), iCAD (July 2025, the ProFound Breast Health Suite), CIMAR UK (cloud image management, November 2025) and Gleamer (March 2026; musculoskeletal, chest and neuro radiograph AI, which keeps its own brand and site). The portfolio runs on DeepHealth OS and divides into enterprise imaging (Diagnostic Suite, Operations Suite, TechLive remote scanning, Reporting Pro) and population health clinical AI (the Breast Suite with SmartMammo Dx, also known as Saige-Dx, with density, risk and breast arterial calcification modules, plus chest, prostate, neuro and thyroid suites). Ten FDA clearances are listed under DeepHealth, Inc., and Chest XRay was cleared in September 2026; the company counts 26 US clearances and 22 CE markings across the combined portfolio. The ASSURE study (Nature Health, November 2025) reported a 21.6 percent increase in cancer detection with the AI driven breast screening workflow across 579,583 women at more than 100 community sites, against a historical comparison. The company reports more than 2,700 customer contracts in more than 50 countries including Gleamer, and names ONRAD and Wichita Radiological Group among third party customers.

Ultromics, a University of Oxford spin out founded in 2017, builds AI that reads a routine echocardiogram and detects disease a reader does not otherwise extract from it. Its EchoGo products work from a single standard view and aim at identifying disease rather than automating measurement. It holds six FDA 510(k) clearances. EchoGo Heart Failure (K222463 in 2022, version 2.0 as K240013 in 2024) aids detection of heart failure with preserved ejection fraction in adults over 25, from an apical four chamber view. It uses a convolutional neural network, returns a result with a 0 to 100 percent score, and declines to classify inadequate studies. The 2024 summary reports 90.3 percent sensitivity and 86.1 percent specificity in a retrospective case control study of 1,578 patients, or 84.9 and 78.6 percent when the 7.4 percent of studies it declined are counted. EchoGo Amyloidosis (K240860, November 2024) is cleared as a screening aid for cardiac amyloidosis, with 84.5 percent sensitivity and 89.7 percent specificity stated by the company. A multi site study in the European Heart Journal, run with Mayo Clinic and University of Chicago investigators, validated it on an external cohort of 2,719 patients, reporting 85 percent sensitivity and 93 percent specificity. Earlier clearances cover EchoGo Core, automated measurement, and EchoGo Pro. The company states that both disease detection products carry Breakthrough Device designation. The service is hosted, and results return into existing clinical systems. EchoGo Heart Failure is billed under CPT Category III code 0932T with payment under APC 5743. The company names Mayo Clinic, City of Hope, University Hospitals and UChicago Medicine as users, and runs a trust center listing ISO/IEC 27001:2022, a HIPAA report and a penetration test report, each on request. Two things a buyer should weigh. The company's own charge is not published, only the payer's code. And where studies are processed, how long they are kept, and whether they feed model development are not stated.

ScreenPoint Medical, founded in 2014 by Nico Karssemeijer and Sir Michael Brady and based in Nijmegen in the Netherlands, makes Transpara, breast AI for mammography and tomosynthesis. Transpara Detection scores each exam from 1 to 10 and in low, intermediate and elevated risk categories and marks suspicious regions; Transpara Density assesses breast density; Temporal Comparison checks findings against up to three prior exams. Transpara 2.1.0 is FDA cleared as a concurrent reading aid (K241831, November 2024) and CE certified. Its evidence centers on MASAI, a randomized trial of more than 105,000 women in the Swedish screening program led by Lund University, which reported a 44 percent reduction in reading workload, 29 percent more cancers detected with no rise in false positives, and a 12 percent lower interval cancer rate (The Lancet Oncology 2023, The Lancet Digital Health 2025, The Lancet 2026). A prospective Spanish study (Nature Medicine, 2026) excluded low risk exams from human review, and a US randomized trial, PRISM, has selected Transpara. The company reports use in more than 30 countries and over 13 million mammograms processed, sold to screening programs, hospitals and imaging groups directly and through partners. In April 2026 it raised $14 million from Insight Partners and Siemens Healthineers.

Copenhagen company, founded in 2018, whose Apollo software reads brain MRI while the scan is still running. Apollo Acquire flags acute infarcts, intracranial hemorrhage and tumors from the first images, alerts the care team, and recommends protocol changes that staff accept in one click, so extra sequences can be acquired before the patient leaves. Apollo Interpret ranks finished studies by urgency across scanners and sites and highlights key images. It is sold to hospitals, private radiology centers, stroke centers and teleradiology providers, directly, embedded in Siemens Healthineers MRI systems, and through AI platforms and distributors. Apollo 2.1.2 is a Class IIa medical device in the EU under the Medical Device Directive, covered by the MDR's transition provision. It is for investigation use only in the United States. Cerebriu's protocoling results give first stage sensitivity of 90.1 percent for infarct, 76.6 percent for hemorrhage and 62.5 percent for tumor, without case counts. The company says it is deployed across Europe and the Middle East. No price, security certification or service privacy document is published.

RapidAI (iSchemaView, Inc.) analyzes brain, vascular and body imaging to support urgent treatment decisions, principally in stroke. Founded in 2012 by stroke neurologist Greg Albers and based in San Mateo, California, it reports use in more than 2,500 hospitals worldwide. Its perfusion software was the patient selection tool in the DEFUSE 3 and DAWN trials that extended the stroke treatment window to 24 hours in US guidelines, and the company states it holds the only FDA indication for perfusion software in selecting patients for thrombectomy. Cleared modules include Rapid LVO, ICH, ASPECTS, NCCT Stroke and aneurysm triage, with five more (DeltaFuse, LMVO, midline shift, obstructive hydrocephalus and aortic) announced in November 2025, plus pulmonary embolism and Lumina 3D reconstruction. They run on the Rapid Enterprise Platform: Rapid EdgeIQ orchestration, Rapid Edge Cloud with an on-premise server and fallback, Navigator Pro for radiologists (worklist prioritization, AI review, in-PACS communication and automated reporting), the Rapid mobile app for care teams, analytics, and partner algorithms. The company states SOC 2 Type 2, ISO 27001, ISO 27701, ISO 42001, ISO 13485 and CSA STAR certification. Pricing is not published.

Eyenuk makes EyeArt, autonomous AI software that grades retinal photographs for diabetic retinopathy and returns a result for each eye without a clinician reading the images. The company is based in San Francisco. EyeArt holds two FDA 510(k) clearances under the retinal diagnostic software classification: K200667 (August 2020) and K223357 for version 2.2.0 (June 2023). The cleared indication is adults with diabetes not previously diagnosed with more than mild diabetic retinopathy, on the Canon CR-2 AF, Canon CR-2 Plus AF and Topcon NW400 cameras, and for each eye the software reports more than mild and vision threatening disease as negative, positive or ungradable. Images are sent from the clinic to analysis servers in a hosted data center. The company describes EyeArt as the first and only FDA 510(k) cleared autonomous AI that detects vision threatening diabetic retinopathy, and reports Class IIb CE marking under the European medical device regulation for version 3.0, covering diabetic retinopathy, age related macular degeneration and glaucomatous optic nerve damage, as well as Health Canada approval. The K223357 summary reports a prospective study at six sites with 171 participants and 336 eyes, graded against a reading center reference: for more than mild disease, sensitivity of 95.9 percent and specificity of 86.4 percent on the Canon cameras and 94.4 and 91.1 percent on the Topcon, with about 90 percent of eyes gradable without dilation. The company states SOC 2 Type 2 attestation, HIPAA and GDPR compliance and ISO 13485 certification, and that screening is billed in the United States under CPT code 92229. It reports that Norway selected EyeArt in 2025 for its national diabetic eye screening program, with results written to the national record. Two things a buyer should weigh. The company's own charge is not published, only the payer's code. And the clearance summary reports no results by race, ethnicity, age or sex, and the company does not say where its data center sits or how long screening images are kept.

Oatmeal Health, founded in May 2022 by Jonathan Govette and based in San Jose, runs lung cancer screening programs for federally qualified health centers and the imaging centers and hospitals that scan their patients, serving Medicaid and Medicare populations that screening has historically reached least. LungIQ reads FQHC record exports, using a large language model and a classifier to recover smoking history and risk factors from clinical notes, and produces a ranked worklist of patients likely to meet the USPSTF screening criteria. A consortium model connects FQHCs without CT scanners to partner imaging centers. LungAI, a vision transformer trained on National Lung Screening Trial data, scores nodule malignancy from 0 to 100 inside the imaging center's existing CADe viewer, with Coreline Soft among the CADe partners. The company is precise about regulatory status: LungAI is an investigational device, not FDA cleared and not commercially available in the United States, with a 510(k) pre submission filed in March 2026 and a pivotal study at Mass General Brigham launching in the second quarter of 2026. Its business associate agreement is published as a click through on its site. What a buyer should weigh: the performance figures (0.96 area under the curve for both models) come from the company's own retrospective validation with no published report; the language model that reads clinical notes is not identified; and no price or contracting model is published, only the reimbursement routes buyers can bill.

Magentiq Eye is a Haifa company founded in 2014 by Dror Zur that builds real time computer aided detection for colonoscopy. Its product, MAGENTIQ-COLO (registered with the FDA as the ME-APDS automatic polyp detection system), takes the video feed from the colonoscope, analyzes each frame with a deep learning model during the procedure, and returns the video to the monitor with suspected polyps marked, so the endoscopist walks past fewer adenomas. It carries a CE mark and Israeli approval from mid 2021 and four FDA 510(k)s under the gastrointestinal lesion detection classification: K223473 (July 2023), K244023 (January 2025), K252178 (October 2025) and K260724 (April 2026), with endoscopic tool detection added along the way. A separately CE marked feature set covers polyp size estimation and characterization plus procedural quality metrics that the US version does not include. The company distributes through an expanded partnership with PENTAX Medical across EMEA, JAPAC, Latin America and North America. Its evidence is a randomized, multicenter tandem trial across ten centers in Europe, the United States and Israel, with about 950 patients and 31 endoscopists, published in Lancet Digital Health in February 2024: adenomas per colonoscopy rose from 0.51 to 0.70 with the device, with a lower adenoma miss rate. The trial is company sponsored, and adenoma detection is a surrogate for cancer outcomes. Two things a buyer should weigh. The site publishes no privacy policy, security information or price. And whether the unit stores, uploads or trains on procedure video is not stated.

Digital Diagnostics, founded in Coralville, Iowa as IDx Technologies by retina specialist and computer engineer Michael Abramoff, MD, PhD, builds autonomous AI diagnostic systems. Its flagship, LumineticsCore (cleared as IDx-DR), detects more than mild diabetic retinopathy including macular edema from retinal images and returns a result at the point of care without physician interpretation of the image: refer to an eye care professional, rescreen in 12 months, or insufficient image quality. In April 2018 it became the first autonomous AI diagnostic system authorized by the FDA, through De Novo (DEN180001) after Breakthrough Device review, followed by 510(k)s in 2021 (K203629) and 2022 (IDx-DR v2.3, K213037) for adults with diabetes on the Topcon NW400 camera. The 2022 summary reports sensitivity of 87.7 percent and specificity of 90.1 percent on the pivotal study data. Images are captured in the clinic and analyzed on hosted servers. The company also lists SphereDx, a diagnostic results platform, and states that DermSpot, a skin product, is not currently authorized or available for sale. LumineticsCore is reimbursable under CPT code 92229, the first national Medicare payment for an FDA authorized autonomous AI service, and the company sells only to healthcare providers. It operates LumineticsCore as an Independent Diagnostic Testing Facility and publishes a notice of privacy practices, which states that patient data is used to improve its AI algorithm. Two things a buyer should weigh. The company's own charge is not published, only the payer's. And no security attestation is published for a service that sends every retinal image, a biometric identifier, to a hosted datacenter.

TeraRecon, a ConcertAI company, sells advanced visualization and AI workflow software for radiology and cardiology. TeraRecon AV (registered as Intuition outside the US) is vendor neutral 2D, 3D and 4D visualization, measurement and reporting across CT, MR and other modalities, with modules for cardiac, structural heart, perfusion and neuro work. The AI Framework, built on the Eureka Clinical AI platform, runs TeraRecon's own algorithms, third party algorithms and a site's in house models through one PACS integration. Physicians review and accept or reject AI results before they reach the report. TeraRecon's own AI devices include Aorta.CT, Cardiac.Chambers.MR and Cardiovascular.Calcification.CT, which segment anatomy and calcium on CT and MR. TeraRecon says more than 2,000 hospitals use its software and publishes indications for use, IHE, DICOM and HL7 conformance statements. It announced ISO 27001:2022 certification in 2023. Headquarters are in Fremont, California. No price, performance figures for its own algorithms, or current list of third party partners are published. ConcertAI's real world evidence business is a separate product and not assessed here.

MEDICAL IP builds AI segmentation and quantification software for medical imaging. MEDIP PRO converts CT and MRI into segmented 3D anatomical models for planning and 3D printing; DeepCatch measures muscle, visceral fat and bone density on CT, and the DeepCatch X modules take measurements from a single chest X-ray (aorta, heart, torso, tuberculosis and emphysema); DeepFore predicts liver cancer occurrence and DeepFore Recur its recurrence; TiSepX performs lung volumetry. The same segmentation feeds a 3D printed anatomical model service (ANATDEL) and anatomy education products, including virtual reality content. MEDIP PRO (2019) and DeepCatch (2023) are FDA cleared, and the company reports Korean approvals up to Class III (DeepFore), CE marking under the Medical Device Directive, UK MHRA registration and a Chinese NMPA certificate for MEDIP PRO. DeepFore was published in the Journal of Hepatology in 2025. Founded in September 2015 as a spin off from Seoul National University Hospital, led by chief executive Sang Joon Park, with offices in Seoul, San Jose and Xi'an.

MediView XR builds XR90, an augmented reality visualization and navigation system for image guided interventional procedures, and markets OmnifyXR for interventional suites. XR90 renders a patient's own anatomy as a 3D hologram derived from their CT, projects live ultrasound into the body in anatomical position as the clinician scans, and tracks the needle path with a holographic guide, so the operator can look at the procedural site rather than away at a monitor. The engineering problem is spatially registering prior CT to live ultrasound and to the instrument in real time, using a commercial off the shelf headset (Microsoft HoloLens 2), a streamer, a GE Vivid iq ultrasound interface and electromagnetic tracking. Founded in 2017 in Cleveland on intellectual property developed at Cleveland Clinic, led by president and chief executive Mina Fahim. XR90 received FDA 510(k) clearance in July 2023 (K223125) for 3D visualization and interaction in conjunction with traditional imaging, for pre operative planning and heads up display during ultrasound guided needle procedures such as biopsies and tumor ablations; the company describes it as the first such clearance for an augmented reality device. The summary reports registration error of about 2.3 to 2.9 mm in phantom, cadaver and animal testing. First clinical use followed in November 2023, a mediastinal lymph node biopsy by Dr Bradley B. Pua at NewYork-Presbyterian / Weill Cornell Medical Center. In October 2025 the company closed a $24 million Series A led by GE HealthCare with participation from Mayo Clinic and Cleveland Clinic. Two things a buyer should weigh. The company's news has described its direction as augmented reality and AI, while no learned component is described in the product or its clearance. And the system depends on a single third party headset line, with remote collaboration sending procedural imaging outside the room under terms that are not published.

Circle Cardiovascular Imaging, based in Calgary and working in cardiovascular imaging since 2007, makes cvi42, vendor-neutral software for reading and reporting cardiac MR, cardiac and vascular CT, structural heart and electrophysiology planning, with StrokeSENS for acute stroke. The cvi42 Coronary Plaque Software Application, cleared by the FDA in October 2025 (K251027), uses semi-automated machine learning to segment coronary vessels and lumen and quantify total, calcified, noncalcified and low attenuation plaque on adults' cardiac CT; clinicians review and edit the contours, and version 6.5 (August 2026) made plaque segmentation editable. The software runs on premise, on a hospital workstation or customer managed server, so coronary analysis happens without sending patient data outside the institution, and the models are locked and do not learn from customer data. Circle publishes its FDA clearances by number and registrations in about twenty other jurisdictions, and a coding and reimbursement guide for plaque analysis. Pricing is not published.

Artrya, listed on the ASX (AYA) and based in Perth, sells Salix, a cloud platform for reading coronary CT angiography at the point of care. Physicians start from an automated analysis, returned in minutes without an external reading team, and edit it before reporting. Machine learning handles centerline extraction, vessel labeling, lumen and wall segmentation and calcium scoring, and plaque is quantified from Hounsfield unit ranges, including low attenuation plaque, spotty calcification, positive remodeling and the napkin-ring sign. The Salix reading platform (K243038) was cleared in March 2025 and Salix Coronary Plaque (K251837) in August 2025; Coronary Flow, an FFR-CT module, is in validation ahead of a 510(k) submission. Salix runs on Amazon Web Services. US customers pay a monthly subscription for the reading platform, published at US$0.6 million over five years for Tanner Health's five hospitals, plus a fee per scan for plaque analysis; Northeast Georgia Health System and Cone Health are further customers. The company markets Salix for rapid assessment of chest pain in emergency and primary care; its SAPPHIRE study across six US health systems, focused on women, has not yet reported.

Heartflow created non-invasive coronary physiology from CT. Its Heartflow One platform analyzes a coronary CT angiogram a hospital has already acquired: Roadmap Analysis for stenosis, FFRCT Analysis for the physiological significance of each lesion, simulated from a 3D model of the coronary arteries, Plaque Analysis and plaque staging, and PCI Navigator for planning intervention, now in routine use at The Valley Hospital. FFRCT was first authorized by the FDA in 2014 and is supported by the ACC/AHA chest pain guideline; Plaque Analysis and Roadmap were cleared in 2022, with further clearances reported in May 2026. Outside the US, Europe, the UK, Australia, Canada and Japan have FFRCT and planning only. FFRCT was validated against invasive FFR in the prospective, multicenter HeartFlowNXT study, and the evidence also includes the PRECISE randomized trial, the ADVANCE registry and the DECIDE registry of about 22,000 patients; the company reports more than 600 publications. Studies are processed in Heartflow's cloud on Amazon Web Services, with trained Case Analysts working on each model, and results return to PACS and the record. The company holds HITRUST, ISO 27001, SOC 2 Type 2 and ISO 13485 certification. Nasdaq listed (HTFL), founded in 2007 and based in Mountain View, California. Pricing is not published; the company documents the reimbursement pathways in detail.

Caristo Diagnostics, an Oxford spinout, measures coronary inflammation from the same routine coronary CT angiogram used to assess plaque. Inflammation changes the fat around the coronary arteries in a way CT can detect; CaRi-Heart uses AI to quantify it as the FAI-Score and to estimate 10 year cardiac mortality risk (the CaRi-Heart Risk score). CaRi-Heart received FDA De Novo authorization in July 2026 (DEN250042), with US commercial launch starting in the third quarter of 2026 and Category III CPT codes 0992T and 0993T from January 2026; it is also available in Europe, Switzerland, the UK and Australia. CaRi-Plaque, for plaque and stenosis quantification, was cleared by the FDA in February 2025 (K242240). The evidence for the inflammation score comes from the Oxford group that founded the company: the ORFAN multicentre cohort in routine NHS care (The Lancet, 2024) and CRISP-CT (The Lancet, 2018) link higher FAI-Score to cardiac events, including in patients without obstructive disease. Analysis runs as a cloud service with trained operators, and the Instructions for Use are published openly. Based in Oxford, England, with a US presence in Stamford, Connecticut. Pricing is not published.

Elucid analyzes coronary and carotid CT angiography to quantify and classify atherosclerotic plaque, including lipid rich necrotic core, the plaque type most linked to heart attack and stroke. Plaque-IQ, cleared by the FDA in September 2024 (K241524), classifies plaque components against ground-truth histology rather than fixed Hounsfield unit ranges, an approach the company calls CT Virtual Histology, and reports patient, vessel and lesion level composition, stenosis and CAD-RADS scoring; the carotid application launched in October 2025. The service is cloud based: Elucid analysts segment each study, the models classify it, and physicians review an interactive viewer and a PDF report and can request rework. An FFR-CT product derived from the plaque analysis, and the stenosis output, are pending FDA review. Plaque-IQ is billable under Category I CPT 75577 from January 2026, with coverage from four of seven Medicare contractors, EviCore and UnitedHealthcare; the company says 70 percent of insured US lives have positive coverage policies. Elucid Bioimaging is based in Boston, with earlier clearances for vascuCAP and ElucidVivo. Pricing is not published.

Beijing based imaging AI company whose InferRead line detects findings on CT and radiographs (lung nodules on chest CT, chest abnormalities on radiographs, coronary stenosis on CTA, hemorrhage on head CT, chest fractures and pneumonia), alongside the InferOperate suite of 3D reconstructions for thoracic, liver and urologic surgical planning, InferCare for patient follow up and RECIST tumor assessment, and the InferScholar research and InferMatrix application platforms. Six US clearances cover InferRead Lung CT.AI (2020 and 2025, nodules of 4 mm and larger in asymptomatic adults aged 55 and over), InferRead CT Stroke.AI (2021, triage), the InferOperate Suite (2025 and 2026) and InferCare RECIST (2026). The lung product runs on premises or in the cloud. Reported deployments include a US partnership with SimonMed and tuberculosis screening programs procured through international agencies in Eastern Europe, Central Asia and Africa.

UK company whose red dot platform sits at the autonomous end of imaging AI: it reports chest X-rays it classifies as High Confidence Normal autonomously, removing them from the radiologist worklist (about 15 percent of studies, according to the company), triages suspected lung cancer for urgent reporting, and has a companion CT head product. Work published in Clinical Radiology reported a 0.33 percent error rate on the auto reported studies and a 60 percent reduction in missed lung cancers when triage ran alongside consultant radiologists. The only FDA clearance, from 2020, is for pneumothorax triage and was granted to the original company. That company, Behold.ai Technologies Limited, entered administration in January 2025 after failing to win contracts under the NHS AI Diagnostic Fund, and NICE withdrew red dot from its recommendations in February 2025. A successor, Behold.ai Global Technologies Limited, acquired the intellectual property and assets and relaunched the platform, positioning it as an alternative to outsourced reporting for the NHS and the US.

Romanian imaging AI company based in Timisoara whose Rayscape CXR analyzes chest X-rays, detecting and localizing 148 findings across 17 clinical classes, including tuberculosis, with bone suppressed views, an automatic cardiothoracic ratio and a normal or abnormal flag for worklist prioritization, and whose lung CT product detects nodules and compares them against a patient's prior scans. Its tuberculosis detection was evaluated independently in a South African tuberculosis prevalence survey published in The Lancet Digital Health in 2024. The products are CE marked under the EU Medical Device Regulation and have no FDA clearance. Rayscape CXR integrates with on premises or cloud archives and returns annotated images inside the radiology viewer, with a reporting engine that drafts report text for radiologist validation; it is also listed on AWS Marketplace and in the Blackford Analysis portfolio. The company reports clinical use at more than 300 sites in 18 countries.

French imaging AI company based in Paris whose Milvue Suite reads plain radiographs. TechCare Alert detects fractures, dislocations, joint effusions, pneumothorax and nodules; TechCare Metrics returns about 60 automated bone measurements, including lower limbs, EOS, bone age and Cobb angles; and TechCare Report pre-fills the radiology report, which the company labels as not a medical device, with results returned in under 30 seconds. In the United States it holds two FDA clearances: TechCare Trauma (K242171, January 2025) for fractures from birth onward and elbow effusions, and SmartChest (K232410, May 2024) for chest X-ray triage. The suite is CE Class IIa under the EU Medical Device Regulation through BSI, with Health Canada Class II licenses for detection and measurements. The software runs on premises or in the cloud. The company reports more than 600 sites in more than 25 countries and 40 million analyses a year.

Paris based imaging AI company whose Rayvolve suite reads plain radiographs: AZtrauma detects fractures, dislocations and joint effusions, AZchest covers chest findings including pneumothorax, pleural effusion and lung nodules, and AZmeasure and AZboneage add measurements and bone age. A second suite, Rayscan, detects lung nodules on CT. AZmed holds six FDA clearances, from fracture detection in 2022 (K220164) through the 2026 extension to dislocations, effusions, spine and ribs for adults and children aged two and over (K260378) and a 2026 AZchest clearance (K261378); the company counts eight FDA cleared findings. Its products are CE marked and its quality system is ISO 13485 and MDSAP certified. The company reports use at more than 2,500 sites in more than 55 countries and 15.4 million X-rays analyzed in the last twelve months, and its clearances rested on retrospective studies at University Hospitals in Cleveland and SimonMed Imaging.

Imaging AI business of Nasdaq listed Nano-X Imaging, formerly Zebra Medical Vision, acquired in November 2021. Its solutions analyze CT scans already acquired for other reasons to surface undiagnosed chronic disease: HealthOST for vertebral height loss and bone mineral density, HealthCCSng for coronary artery calcium on non gated chest CT, and HealthFLD for liver density in fatty liver disease. The company states seven FDA clearances and two CE marks; the first bone clearance, HealthVCF, dates from 2020. Named users include Corewell Health, Atlantic Health System, Jefferson Einstein Hospital, Rabin Medical Center and NHS hospitals, and reported customers include integrated delivery networks, health plans and pharmaceutical companies.

VUNO, a Seoul medical AI company founded in 2014, leads with DeepCARS, which predicts in-hospital cardiac arrest within 24 hours for patients on general wards from four routinely charted vital signs: blood pressure, heart rate, respiratory rate and temperature. It returns a 0 to 100 risk score for ward staff and rapid response teams to act on, and the company describes it as Korea's first AI medical device for this purpose. DeepCARS is approved by Korea's Ministry of Food and Drug Safety, carries a CE mark and holds FDA Breakthrough Device designation; no US clearance for it appears in the FDA 510(k) database. Its evidence includes a multicenter validation of 173,368 patients published in Resuscitation, reporting an area under the curve of 0.905 on external validation, and a prospective multicenter study in Critical Care, and the company reported in June 2026 a study in secondary hospitals finding 21 percent fewer cardiac arrests. The company has reported deployment across more than 48,000 beds in South Korea, where since 2022 hospitals can charge patients for it as a non-covered service under the early access medical technology route. VUNO also sells imaging AI, which places it among imaging vendors: VUNO Med-Chest X-ray, cleared by the FDA for triage (K241439, November 2024), and VUNO Med-Fundus AI. It holds a second US clearance for VUNO Med-DeepBrain, brain MRI quantification (K231398, October 2023), and sells in the United States through a subsidiary, VUNO MED Inc. Its privacy policy states that it processes patient data on behalf of hospitals, with retention periods by product, named processors including AWS, no overseas transfer, and use of pseudonymized data for AI development. It holds ISO/IEC 27001 and ISO/IEC 27701 certification through 2028. Two things a buyer should weigh. Accuracy for DeepCARS is published as discrimination rather than alert and miss rates at an operating threshold. And the published privacy and residency terms describe the Korean service; the US subsidiary's HIPAA posture is not stated.

Ohio based company whose ClearRead software applies AI to chest CT and chest radiographs. ClearRead CT produces a vessel suppressed series with nodule detection marks, cleared by the FDA in 2016 (K161201) as an aid to the radiologist for pulmonary nodules on chest CT in an asymptomatic population, and extends to coronary artery calcium scoring (K242188, 2024). ClearRead Xray covers bone suppression and nodule and pneumothorax detection, with pneumothorax triage cleared in 2022 (K213566). The company reports more than 400 customer locations.

Cleerly applies AI to coronary CT angiography to quantify and characterize atherosclerotic plaque and estimate ischemia without an invasive test. Cleerly Labs, web based analysis cleared in 2020, measures plaque volume, composition, length and location and stenosis on CCTA from any scanner; Cleerly ISCHEMIA, cleared in 2023 (K231335), estimates vessel level ischemia as a diagnostic aid; Cleerly COMPARE tracks change between scans. ISCHEMIA was validated in the prospective, multicenter CREDENCE trial against invasive FFR, with further support from a PACIFIC post-hoc analysis and the international CONFIRM2 registry. Reimbursement is the company's commercial strength: Category I CPT codes for plaque analysis (75577, from January 2026) and ISCHEMIA (75580), Medicare contractor coverage and commercial payer policies reported at more than 86 million covered lives, with a public payer policy lookup tool. Services are delivered from the United States; a public trust center offers SOC 2 Type 2 and HITRUST reports for download. Founded in 2017 and based in Denver, Colorado. Per study pricing is not published.

Radiology AI company headquartered in Sydney. Its radiology products were sold as Annalise.ai, including Annalise Enterprise CXR and Annalise Enterprise CTB, until Harrison.ai combined with Annalise.ai in February 2025 and renamed the Annalise.ai solutions Harrison.ai Radiology in October 2025; the annalise.ai domain now forwards to harrison.ai. It sells detection software for chest X-ray (up to 124 findings), non contrast head CT (up to 130 findings) and chest CT (up to 167 radiological features) as clinical decision support for radiologists and other clinicians, alongside a radiology foundation model, Harrison.Rad 1.5. In the United States its FDA clearances are for triage and notification of specific findings on chest X-ray and head CT, filed as Annalise Enterprise devices; the comprehensive detection products are sold under CE marking and other regimes outside the US. The company reports that more than 35 percent of chest X-rays in NHS England are processed with its chest X-ray product.

Gleamer SAS is a radiology AI company based in Saint-Mandé, on the edge of Paris, and since 2 March 2026 part of DeepHealth, the digital health subsidiary of the US imaging services operator RadNet, Inc. RadNet announced the all cash acquisition at up to 230 million euros, about 215 million at closing plus a 15 million contingent milestone. Gleamer's site and products keep the Gleamer name. Its lead product, BoneView, detects fractures on radiographs as a concurrent reading aid and is FDA cleared for adults across all anatomical regions and for children and adolescents aged 2 to 21 on the limbs (K222176, 2023). The Gleamer Copilot suite also includes ChestView for chest X-ray and BoneMetrics for musculoskeletal measurements, both FDA cleared, and BoneAge, LungCT, BoneCT, BreastView, the Pixyl neuro products for multiple sclerosis and brain volumetry, LumbarMR, and two reporting products, Voice and AutoReport, for which no US clearance under Gleamer is listed. The company reports more than 60 million exams a year across more than 3,500 clinical sites in more than 50 countries. At acquisition it reported more than 130 staff and more than 700 customer contracts, and RadNet stated it intends to deploy the tools across its own network of more than 400 imaging centers.

Lithuanian radiology AI developer based in Vilnius, part of Sectra, the enterprise imaging company, since 2026. Its ChestLink product, which the company describes as the world's first autonomous AI medical device, holds CE Class IIb certification to identify normal chest X-rays and issue final reports on them with no radiologist involved; the company states that autonomous operation is available in Europe and not in the United States, and no FDA clearance is listed for any of its products. The CXR Suite adds ChestEye notifications on 75 chest X-ray findings for radiologists and a quality review module, and the CE certified CT Eye and MSK Eye extend detection to chest CT and musculoskeletal X-ray. The company posts its EU device, ISO 13485 and ISO 27001 certificates on its website.

AI powered care coordination platform, Viz One, that pairs detection algorithms on imaging and ECG with automatic alerts and team messaging on clinicians' phones. It began in stroke: its large vessel occlusion product, ContaCT, received FDA's De Novo authorization in 2018 (DEN170073), which created the regulatory category for imaging triage and notification software. The company now reports more than 50 FDA cleared algorithms, sold as suites for neurovascular, vascular, cardiac, pulmonary, trauma and radiology care, with oncology and other areas offered as pilot programs, plus a workflow assistant, Viz Assist. The stroke product has the strongest evidence, including a multicentre prospective randomized trial that reported shorter door to groin times. Viz.ai states SOC 2 Type II, ISO 27001 and ISO 42001 on its Trust Center, with reports available on request, and runs on Amazon cloud infrastructure across about 2,000 hospitals and health systems in the United States and Europe. It also sells to life sciences companies. No price is published.

Cancer focused medical AI company founded in 2013, headquartered in Seoul and listed on KOSDAQ. Its radiology line is Lunit INSIGHT: MMG for 2D mammography (FDA cleared 2021), DBT for tomosynthesis (cleared 2023) and CXR for chest X-ray (CXR Triage cleared 2021). Lunit acquired Volpara Health Technologies of Wellington, New Zealand, in May 2024; the Volpara name was retired in November 2025 and the business became Lunit International, whose density, risk, quality and patient outreach products (Scorecard, Risk Pathways, Analytics, Live, Patient Hub) now sit alongside INSIGHT in a breast cancer platform. A separate line, Lunit SCOPE, applies AI to pathology for biopharma research, companion diagnostic development and pathologists under research use; because it serves different buyers, it is covered by a separate record, Lunit SCOPE. The company reports more than 10,000 customer sites in over 70 countries and more than 700 publications, and sells direct and through imaging equipment partners. The strongest evidence is ScreenTrustCAD, a prospective Swedish screening study of 55,581 women in which AI with one radiologist was non-inferior to two radiologists; an Emory led evaluation with Lunit co-authors reports INSIGHT DBT performance by patient subgroup. In the US, INSIGHT MMG is cleared as an adjunct viewed after the radiologist's initial read, and SecondRead AI offers it to patients as a paid add-on. A buyer searching for Volpara or Volpara Health is searching for this vendor.

Radiology AI company founded in 2016 in Mumbai, with a US entity registered in New York, selling in more than 105 countries through more than 5,500 sites by its own count. Products: qXR for chest X-ray (detection, reporting, lung nodules, tuberculosis screening), qER for head CT in stroke and trauma, qCT for chest CT nodule measurement, qMSK for musculoskeletal X-ray, and heart failure detection. Around them sit qTrack for tuberculosis care cascades, a lung nodule management pathway, a stroke care coordination suite for hub and spoke networks, and Aira, a care delivery platform for health programs deployed in Nigeria, Kenya, Mozambique and the Solomon Islands. In the US, qXR-Detect was cleared in January 2026 (K251934) as a concurrent reading aid across six chest X-ray categories with a Predetermined Change Control Plan, which the company describes as the only one for a chest X-ray detection device; qXR-LN (K231805) and qER are among its other clearances, which the company counts at 26 indications across nine products. Its tuberculosis products are WHO evaluated and independently studied in several countries. Qure says its cloud path receives only de-identified images, runs on AWS and Azure, and holds ISO 27001 certification. Pricing is not published.

Company founded by interventional radiologists to find peripheral arterial disease in imaging already taken. Its one product in market, VascularAssist Occlusion Triage (VAOT), was cleared through FDA 510(k) K260729 in June 2026 as Class II triage and notification software. It reads adult CT angiography and notifies the radiologist through the PACS when it finds a pattern consistent with total occlusion between the aorta and the popliteal arteries. It does not diagnose, and the radiologist still reads every study. In the company's pivotal study of 317 cases, sensitivity was 91.5 percent and specificity 90.1 percent. A gateway at the hospital de-identifies each study before cloud analysis, and an on premises option is offered. The website also describes Elevate, a workflow layer to route and track patients after a finding, plus plaque characterization, carotid and coronary modules, and an executive health offer. The company labels all of these as in development or investigational and not cleared. It sells through teleradiology (a model development agreement with vRad), academic centers, and hospitals and specialty practices, the last on a platform fee plus a per study charge. The listed parent, GuideAI Health Corp., is a British Columbia company trading on Cboe Canada as GDAI since June 2026; the operating company is in Boston.

Seoul based MRI AI company with US operations. SwiftMR is a deep learning enhancement tool that reduces noise and sharpens MRI images after acquisition, so sites can scan faster or get better images from the same scan. It works across scanners from the major manufacturers and field strengths from 0.25 to 3.0 tesla, and now alongside scanner makers' own deep learning reconstruction. SwiftSight produces quantitative brain MRI measurements. AIRS Medical reports more than 1,700 institutions in over 40 countries. SwiftMR has been FDA 510(k) cleared since 2021; the latest clearance, K253775 (2026), covers enhancement of all body parts and was tested on image quality rather than diagnostic accuracy. It runs in the cloud, with identity removed inside the hospital before images leave, or on premises, and original and processed series can be archived separately. AIRS Medical names ISO/IEC 42001 AI governance and publishes an AI Transparency Statement and security summaries. No price is published.

Overjet is a dental AI company that analyzes radiographs in real time, detecting and outlining caries, periodontal bone loss, calculus and periapical radiolucency with overlays the dentist reviews chairside. Each function carries its own FDA clearance, eight in all from 2021 to 2026, including caries detection on bitewing and periapical radiographs (K222746), charting, image enhancement, cone beam CT (K251514, December 2025) and the IRIS imaging system (K253930, April 2026). The caries clearance reports standalone sensitivity of 76.6 percent on bitewings and 79.4 percent on periapicals against a three dentist consensus, with lower sensitivity for secondary caries. The platform has grown around the detection models: IRIS imaging software for capturing radiographs, automated insurance eligibility and benefits verification, a Voice AI suite for documentation, and DSO analytics. The company also sells a separate claims review product to dental insurers, so its models inform both sides of a dental claim, which it frames as aligning the objectives of dentists and insurers. It describes itself as the #1 dental AI platform, which is its own claim. Overjet was founded in 2018 by Harvard School of Dental Medicine and MIT alumni and is based in Boston. Its trust center names a HITRUST certification, states that Overjet operates as a business associate, and publishes a subprocessor list that includes Google Cloud Platform and large language model providers OpenAI and Anthropic. It sells to individual practices, dental groups and DSOs, and dental insurers. Two things a buyer should weigh. The practice and payer products are counterparties in claims disputes, and nothing published says what technically and contractually separates them. And no price is published for any product.

New York company, founded in 2024, building deep learning tools for PSMA PET/CT in prostate cancer and theranostics. DeepPSMA detects lesions, assigns TNM stage and quantifies whole body tumor burden; SelectPSMA predicts response to PSMA targeted radioligand therapy; TrackPSMA compares scans over time to evaluate treatment response. Studies are uploaded to a cloud platform on Google Cloud, where a physician reviews and edits the findings before a report is generated. The company says more than half of cases need no manual edits. All three products are investigational: not FDA cleared or CE marked, with both said to be in progress, and not for clinical diagnostic use. Performance figures come from internal studies without case counts (92 percent lesion detection accuracy, 87 percent staging). It sells to clinicians, pharmaceutical companies and researchers, and names Segmed as a data partner. No price or security certification is published.

HOPPR builds infrastructure for making medical imaging AI and, separately, a layer that delivers AI results into radiology reporting. Grace is a multimodal imaging foundation model for X-ray, CT, MRI and echocardiography, trained on more than a petabyte of permission based, anonymized studies paired with reports; narrative models translate chest radiographs (April 2026) and 2D mammograms (August 2026) into structured text and ship with training data traceability records and version locking. The HOPPR AI Foundry combines computing, curated datasets, foundation models, fine tuning tooling and a quality management system framework for developers, AI companies and imaging software vendors, with NVIDIA's open models NV-Reason and NV-Generate available since March 2026, and Forward Deployed Services pairs HOPPR machine learning staff with customer teams. Presto Agent, sold separately, is a model agnostic layer for AI draft reporting that places results from models a practice chooses into its existing reporting application, such as PowerScribe. The Foundry holds SOC 2 Type I and Type II attestations and HITRUST e1 certification. Founded in 2019 and led by Dr Khan Siddiqui; $34.5 million raised, including a $31.5 million Series A in June 2025 with Health2047, the American Medical Association venture studio, among investors.

Iterative Health, legally Iterative Scopes, Inc., is based in Cambridge, Massachusetts and runs two businesses under one name: a cleared AI device for colonoscopy and a clinical research site network. Its website leads with the research network, and the device is one of three solutions listed in the footer. SKOUT is real time computer aided polyp detection software. It reads the colonoscope's video, marks potential polyps with boxes on the screen, and leaves every decision to the gastroenterologist. It received FDA 510(k) clearance K213686 in August 2022 for white light colonoscopy in adults having colorectal cancer screening or surveillance, followed by five special 510(k)s for later versions through December 2025. The clearance summary reports a randomized trial of 1,359 patients and standalone testing on 94 polyps, with a per polyp detection rate of 97.9 percent, 22.6 false marks per 15 minutes, and results by sex, age, race, polyp size and histology. The published trial, in Gastroenterology, reports a 27 percent relative increase in adenomas per colonoscopy. In the United States the device is sold through an exclusive distribution partnership with Provation. The research network owns and operates research sites and embeds research in partner provider organizations, working for pharmaceutical, biotech, device and contract research sponsors. The company reports more than one hundred sites across North America, Europe, India and Australia in gastroenterology, hepatology, obesity and cardiology, and describes using machine learning to pre-screen patients for trial eligibility. Two things a buyer should weigh. The two businesses are separate propositions for different buyers, with different compliance questions. And nothing published covers HIPAA status, a security attestation, or what happens to procedure video and screened records, including whether they inform model development.

Clinical AI for imaging triage and follow-up, delivered through aiOS, an enterprise platform that runs Aidoc's own FDA cleared algorithms alongside vetted partner algorithms and carries care coordination and patient management. Aidoc Medical was established in early 2016 in Tel Aviv and is led by co-founder and chief executive Elad Walach. Its models are built on CARE (Clinical AI Reasoning Engine), a self-developed foundation model: CARE1 launched in December 2024, a successor has been announced, and a multi year collaboration with Amazon Web Services supports the work. On 7 January 2026 the FDA cleared BriefCase-Triage: CARE Multi-triage CT Body (K252970) for triage of 11 acute findings on chest, abdomen and pelvis CT, which the company combines with three earlier cleared findings in a 14 finding workflow. In the company's retrospective multicenter test across 6 US sites, sensitivity ran from 94 to 99 percent and specificity from 96 to 99 percent across the 11 findings. The company reports use in more than 1,600 medical centers and more than 100 million patient cases analyzed on aiOS, sells across neurovascular, vascular, cardiology and body imaging, and plans to extend CARE to all CT and X-ray workflows, with automated draft reporting in development. A white paper with Mercy, commissioned by AVIA, counts more than 50,000 new findings surfaced over five months. Security rests on a claimed ISO 27001 certification and a trust center; business associate status and whether customer data trains the models are not stated publicly.

Generative AI for radiology reporting and follow up. Rad AI Reporting is a cloud native reporting system with real time quality checks before sign off; Rad AI Impressions (Omni Impressions) drafts the report impression from the radiologist's dictated findings, customized to each radiologist's language from the site's de identified historical reports; Omni Unchanged carries stable findings forward from prior reports; Rad AI Continuity tracks follow up recommendations through the EHR and RIS. The company runs its own generative models trained on de identified radiology reports, and a licensed radiologist reviews, edits and signs every report. It reports more than 200 health organizations and 11,000 radiologist users, including AdventHealth, Advocate Health, Emory and Trinity Health, and reports follow up completion rising from about 30 percent to over 75 percent with Continuity. Series C funding reached $68 million, including strategic investments from four health systems.

Deep learning software that enhances and accelerates image acquisition on existing scanners rather than detecting disease. SubtleHD denoises and sharpens MRI (cleared February 2025, K243250, with a Predetermined Change Control Plan), PET (cleared May 2026, supporting all FDA approved radiotracers) and CT (the company's first CT clearance and its eleventh FDA cleared product), succeeding SubtlePET (2018) and SubtleMR. SubtleSYNTH generates synthetic spine STIR images from other sequences and SubtleALIGN aligns brain MRI; with SubtleHD (MR) they form the Subtle-ELITE package. SubtleGAD, for lower gadolinium dosing, is in development, and AiMIFY is sold by Bracco. The company claims up to 80 percent faster MRI and up to 75 percent faster PET, reports deployment on more than 1,300 scanners with customers including Mount Sinai, RadNet and Radiology Partners, and lists prospective multicenter MRI trials among its publications. Studies pass through an on-premise SubtleEDGE gateway that strips identifiers before AWS cloud processing, or run fully on premise; de-identified images are shared back to Subtle by default unless a customer opts out. Holds SOC 2 Type II, ISO 13485 and ISO/IEC 42001 certification. Headquartered in Menlo Park, California.

Dental AI for radiographic pathology detection. Not to be confused with Pearl Health, a separate company in value based care intelligence also indexed here. Second Opinion analyzes bitewing, periapical and panoramic radiographs in real time chairside, highlighting findings such as caries, bone loss, calculus, periapical radiolucencies and defective restoration margins, and Second Opinion 3D extends it to CBCT. Eight FDA 510(k)s cover the family, from the original Second Opinion (K210365, 2022) to the panoramic product (K250525, November 2025), including pediatric, bone level and 3D clearances; the company describes itself as the first dental AI company cleared for both 2D and 3D radiologic analysis and reports more than 26 regulatory clearances across more than 120 countries. The panoramic clearance reports lesion level sensitivity of 77 percent for caries, 82 percent for periapical radiolucency and 99 percent for impacted third molars on 795 radiographs, with a reader study and subgroup checks across six panoramic manufacturers. The platform now spans six products: Second Opinion, Second Opinion 3D, Practice Intelligence (practice analytics), Pearl Voice (ambient voice notes), RCM (insurance claims optimization) and Calibrate (diagnostic standards and training for groups). It integrates with more than 40 dental imaging and practice management systems, runs on Amazon Web Services with a selectable US or EU jurisdiction (EU data stored in Dublin by Pearl EU), and names a SOC 2 Type II report supplied on request. The company reports use at more than 29,000 practices. Founded 2019 by Ophir Tanz, headquartered in West Hollywood, California. Two things a buyer should weigh. The home page sells the product on outcomes (37 percent more disease detected, 30 percent more treatment acceptance, $30,000 more monthly production per practice) with no method stated and no price published. And whether customer radiographs train the models is not stated, while group onboarding ingests up to 18 months of historical images.