Cancer-detection AI for radiologists and pathologists, live in 65+ countries
Lunit is a medical AI software suite for cancer detection across radiology imaging and digital pathology.
AI Panel Score
6 AI reviews
Reviewed
AI Editor ApprovedApproved and published by our AI Editor-in-Chief after full panel analysis.Lunit is a medical AI software company that builds deep-learning tools for cancer detection in radiology and digital pathology. It serves radiologists, breast-screening programs, pathologists, and biopharma teams developing companion diagnostics. Pricing is quote-based and arranged through the vendor's sales team rather than published online. The INSIGHT line analyzes chest X-rays, mammograms, and 3D tomosynthesis, detecting findings such as lung nodules, pneumothorax, and breast lesions while flagging normal studies so readers focus on suspicious cases. The SCOPE line reads pathology slides to quantify PD-L1 and HER2 biomarkers, map the tumor microenvironment, and predict driver mutations like EGFR and KRAS directly from H&E images. INSIGHT tools carry FDA clearances and CE marks and run on-premise or in the cloud with DICOM and PACS integration. Lunit fits hospitals, screening networks, and oncology drug developers, and competes with Qure.ai, Annalise.ai, and Aidoc in radiology, and PathAI and Paige in digital pathology.
Lunit runs deep-learning models against medical images to surface cancer signals a reader might miss. In radiology, its INSIGHT applications overlay detection maps on chest X-rays, 2D mammograms, and 3D tomosynthesis, assign abnormality scores, flag normal studies so radiologists can prioritize suspicious cases, and compare current scans against prior exams to track nodule progression. Deployment runs on-premise or in the cloud, with a DICOM gateway handling anonymization and PACS routing.
INSIGHT CXR detects 11 thoracic findings including nodules, pneumothorax, consolidation, and pleural effusion, and supports tuberculosis screening, while INSIGHT MMG and DBT target breast screening and reduce callback rates. The SCOPE line reads digitized pathology slides: SCOPE IO maps the tumor microenvironment through immune phenotyping, the SCOPE IHC Suite quantifies PD-L1 tumor proportion score and HER2 expression, and SCOPE GP predicts driver mutations such as EGFR, KRAS, and ALK from H&E slides without extra tissue sampling. INSIGHT Manager tracks sensitivity, specificity, and AI output trends across sites.
Radiologists, breast-screening programs, pathologists, and biopharma teams building companion diagnostics use Lunit; INSIGHT tools hold FDA clearances and CE marks and are deployed across more than 10,000 sites in 65+ countries. Pricing is sales-led and quote-based rather than published. It competes with Qure.ai, Annalise.ai, and Aidoc in radiology, and with PathAI, Paige, and Owkin in AI pathology and biomarker analysis.
The Volpara product line adds breast-density scoring, risk pathways, and screening analytics alongside the core suite. Lunit reports AUC accuracy in the 95-100% range for INSIGHT CXR and a 36% radiologist workload reduction from normal-case flagging, with results supported by 700+ peer-reviewed publications.
Monitors application performance, sensitivity, specificity, and AI output trends across deployed clinical sites.
Models individual breast-cancer risk by integrating imaging results with patient data.
Predicts driver mutations such as EGFR, KRAS, and ALK directly from H&E slides without extra tissue sampling.
Scores HER2 expression, including ultralow-expression cases, from IHC-stained tissue.
Quantifies PD-L1 tumor proportion score from immunohistochemistry slides to support immunotherapy decisions.
Anonymizes images and routes studies between hospital PACS and Lunit for on-premise or cloud deployment.
Maps the tumor microenvironment through AI immune phenotyping of H&E-stained pathology slides.
Detects 11 thoracic abnormalities on chest X-rays, including nodules, pneumothorax, and consolidation, and supports tuberculosis screening.
Speeds 3D tomosynthesis reading by highlighting the key slices that contain suspicious findings.
Analyzes 2D mammograms to mark suspicious lesions and reduce callback rates in breast-cancer screening.
Measures breast density from mammograms to support risk assessment and screening decisions.
Identifies studies with no abnormal findings so radiologists can prioritize suspicious cases and cut reading volume.
Pricing requires contacting the vendor.
A public, FDA-cleared cancer-detection AI that consolidated the field rather than merely surviving it.
“Lunit pairs FDA clearances and a 2022 public listing with deployments across 10,000+ sites, a rare depth of proof in medical AI. The catch is enterprise procurement and quote-based pricing, so value depends on a disciplined pilot before commitment.”
FDA clearance is the gate most medical-AI startups never reach. Lunit cleared it, then listed on the KOSDAQ in 2022. That's a rare combination of regulatory proof and public-market scrutiny in a category littered with quiet shutdowns.
The strategic question isn't capability — INSIGHT CXR reading 11 thoracic findings across 10,000+ sites answers that. It's whether cancer-detection AI advances your screening program or just adds a line item. The 36% workload-reduction figure suggests the former, but that's their number, not an independent audit.
Their $193M Volpara acquisition in 2024 tells me they're consolidating, not surviving. Against Aidoc and Qure.ai, Lunit's radiology-plus-pathology span is the differentiator. Run a paid pilot at two reading sites, measure your own callback numbers, before you sign anything enterprise-wide.
Radiology-plus-pathology breadth differentiates it from single-domain rivals like Aidoc.
FDA clearances and 700+ publications make this a defensible board choice.
On-premise enterprise deployment and procurement lengthen time to first result.
Fits screening programs directly, though value hinges on integrating with existing reader workflow.
Public on KOSDAQ since 2022 with a $193M acquisition signals rare staying power.
Screening programs who want regulator-cleared cancer-detection AI at scale.
Small clinics who need published pricing before evaluating.
For a CMO, Lunit's citation depth and dual radiology-pathology span justify a serious clinical bet.
“Lunit backs its INSIGHT and SCOPE lines with 700+ peer-reviewed studies, spanning cancer detection through biomarker-driven treatment selection. The on-premise validation and IT overhead are meaningful, but justified for institutions running screening at population scale.”
Seven hundred peer-reviewed publications is not a vanity metric in oncology — it's the clinical governance asset that survives a procurement committee. For a CMO, that body of evidence is what lets me defend an AI-assisted screening pathway to a tumor board. Lunit built the citation base most imaging-AI vendors never assemble.
The strategic depth is the two-line span. INSIGHT covers radiology screening while SCOPE reads PD-L1 and HER2 for precision oncology, which means one vendor touches both detection and treatment selection. If our institution invests here over three years, we consolidate diagnostic and companion-diagnostic workflows rather than stitching Aidoc to PathAI.
The integration surface is real but heavier than it looks. The DICOM Gateway routes cleanly into PACS, however on-premise oncology deployments carry validation and IT overhead most clinical teams underestimate. For a screening program at population scale, that's an acceptable cost.
Few competitors match this span; Aidoc and PathAI each cover only one side.
INSIGHT and SCOPE map directly onto oncology screening and treatment pathways.
DICOM Gateway routes into PACS cleanly, but on-premise oncology deployment adds overhead.
Consolidating detection and companion diagnostics reduces multi-vendor fragmentation over three years.
700+ publications and dual radiology-pathology coverage give unusual clinical depth.
Cancer centers who want one vendor across detection and biomarkers.
Small radiology groups who lack IT deployment resources.
Zero published prices, but a 36% workload cut gives ROI a rare hard anchor.
“Lunit sells INSIGHT and SCOPE on quote-based enterprise licensing with no public sticker, so year-three cost for a multi-site program reaches seven figures. The 36% workload-reduction claim gives ROI a genuine anchor, but procurement gets no price discipline until sales engages.”
No published prices. In enterprise medical imaging, that's category norm, not a red flag. Lunit licenses INSIGHT and SCOPE per site, per modality, on a quote basis.
Model a mid-size screening program. Ten reading sites, chest and mammography modules, on-premise. Enterprise imaging-AI runs roughly $40K-$80K per site annually, based on category norm. Year three lands in seven figures once storage and IT overhead are counted. The DICOM Gateway is bundled; the on-prem hardware is not.
ROI has an actual anchor here. A 36% radiologist workload reduction converts to real reader hours. The catch is procurement — quote-based deals mean no price discipline until sales engages. Against Aidoc's similar enterprise model, neither wins on transparency, but Lunit's $193M Volpara acquisition signals it'll be there to invoice you in year three.
Enterprise quote-based process fits health systems but slows evaluation.
Per-modality, per-site licensing lets buyers scope to actual deployment.
No published list price; everything runs through a sales quote.
The 36% workload-reduction figure converts to quantifiable reader hours.
Per-site licensing plus on-prem hardware pushes multi-site programs to seven figures.
Health systems who budget medical software through enterprise procurement.
Buyers who need list pricing to compare options.
Normal Case Flagging reorders the worklist, and INSIGHT overlays land inside your existing PACS.
“Lunit's Normal Case Flagging and INSIGHT CXR overlays fit the reading-room worklist and route through PACS via the DICOM Gateway. Workflow integration is strong, but eleven flagged findings per model add per-study decisions that can lift callbacks before they settle.”
The number that matters in a reading room isn't AUC — it's how many normals you clear before lunch. Lunit's Normal Case Flagging triages studies with no findings so the worklist front-loads the suspicious ones. On a high-volume chest screening list, that reorders the whole day.
INSIGHT CXR overlays detection maps and an abnormality score directly on the image, and compares against priors to track nodule change. That's real reading-room value, not a demo trick. But every overlay adds a decision — trust the mark or over-read around it — and 11 flagged findings can push callbacks up before they settle down.
Integration is the strong part. The DICOM Gateway routes results into existing PACS, so marks appear where you already read instead of a second monitor. Against Aidoc's worklist prioritization, Lunit matches on workflow fit and adds pathology its rivals don't touch.
Normal Case Flagging genuinely reorders a high-volume worklist beyond the demo.
INSIGHT Manager exposes sensitivity and specificity trends readers can monitor.
Eleven flagged findings per model can add over-reading and callback pressure.
Abnormality scores, priors comparison, and 11-finding detection give deep reading support.
DICOM Gateway routes marks into existing PACS reading workflow.
Radiologists who read high-volume chest and breast screening lists.
Solo readers who prefer an unmarked image first.
It lands inside your PACS instead of a separate login, and mostly earns its place.
“Lunit's INSIGHT overlays appear inside your existing PACS through the DICOM Gateway, and the normal-flag quietly clears boring studies off a 100-study shift. It's not a mobile tool, and an eleven-finding model occasionally cries wolf, but it feels like part of the read rather than a pop-up.”
The best thing software like this can do is show up where you already work and then get out of the way. Lunit mostly does. The INSIGHT overlay lands inside your existing PACS viewer through the DICOM Gateway, not a separate login you'd forget to open by Thursday.
What sells it day to day is the abnormality score and the normal-flag that clears the boring studies off your list. That adds up over a 100-study shift. But an AI that marks eleven findings also cries wolf now and then, and you learn its personality the slow way — three months in, not three minutes.
This isn't a mobile tool, and that's fine — nobody reads a chest CT on a phone. Reliability feel is high because it's FDA-cleared and monitored through INSIGHT Manager. Against Aidoc, it feels less like a pop-up alert and more like part of the read.
Overlays and abnormality scores land inside the existing PACS viewer cleanly.
Overlays are intuitive, but reading around eleven flagged findings takes weeks.
Not a mobile use case; diagnostic reading happens on workstations.
Enterprise onboarding and clinical support carry setup, not self-serve signup.
FDA clearance and INSIGHT Manager monitoring signal dependable output.
Readers who want AI marks inside their existing viewer.
People who want a lightweight tool with mobile access.
700 publications and a KOSDAQ listing answer the graveyard question most imaging AI can't.
“The AUC 95-100% headline invites skepticism, but 700+ publications, FDA clearances, and a 2022 KOSDAQ listing back it up. Pricing is quote-only so exit math stays hidden, though DICOM-standard images keep your underlying data portable.”
Medical imaging AI has a graveyard. IBM Watson Health promised oncology and got sold for parts. Plenty of imaging startups raised big, then folded quietly. So an AUC 'in the 95-100% range' is the kind of headline number I read twice.
Then the counter-evidence stacks up. 700+ peer-reviewed publications. FDA clearances, not just CE marks. Public on the KOSDAQ since 2022, and a $193M Volpara acquisition — companies about to fold don't buy competitors. Deployed across 10,000+ sites. That's a real track record, not a pitch deck.
Two things I'd watch. Pricing is quote-only, so the exit math is invisible until you're already in. And SCOPE competes with Paige and PathAI in pathology, where Lunit is newer than in radiology. But the images are DICOM-standard, so portability of your data holds even if you leave. Fair for a public 13-year-old vendor.
Dual radiology-pathology span differentiates it from single-domain Aidoc and Paige.
DICOM-standard images stay portable, but quote contracts obscure switching costs.
Public listing since 2022 and $193M M&A point to staying power.
AUC 95-100% is a superlative, though 700+ publications largely substantiate it.
10,000+ sites and FDA clearances match the marketing claims.
Buyers who want a proven public vendor with clinical evidence.
Teams who need transparent pricing before committing.
Common questions answered by our AI research team
Lunit covers lung and thoracic disease through INSIGHT CXR, breast cancer through INSIGHT MMG and DBT mammography, and multiple solid tumors through SCOPE pathology models that read biomarkers like PD-L1, HER2, and driver mutations.
Yes. Lunit INSIGHT CXR and INSIGHT MMG hold FDA 510(k) clearances and CE marks, and the tools are deployed across more than 10,000 clinical sites in over 65 countries with peer-reviewed clinical validation.
Yes. Lunit deploys on-premise or in the cloud and connects to hospital PACS through a DICOM Gateway that anonymizes images and routes studies, so AI results appear inside existing radiology reading workflows.
SCOPE reads digitized H&E and IHC slides: SCOPE IO maps the tumor microenvironment, the SCOPE IHC Suite quantifies PD-L1 and HER2, and SCOPE GP predicts mutations like EGFR and KRAS to support oncology research and companion diagnostics.
Lunit does not publish list prices. It licenses INSIGHT and SCOPE to hospitals, screening programs, and biopharma partners on a quote basis, sized to the modalities, deployment model, and number of sites involved.
Lunit is a Seoul-based medical AI company developing deep-learning software for cancer screening in radiology and biomarker analysis in digital pathology.
