Lunit applies imaging AI to cancer screening and oncology research

Lunit develops AI software for cancer screening and precision oncology research. INSIGHT products support imaging interpretation, while SCOPE products analyze pathology images for research applications such as biomarker development. The company also includes the former Volpara breast-health portfolio. These are related businesses with distinct workflows, and a useful buying decision keeps their intended uses and evidence separate.

01 / ProductScreening assistance and pathology research are two different product families

INSIGHT MMG supports analysis of two-dimensional mammograms, presenting areas of interest and malignancy-likelihood information for professional review. The product page describes PACS and viewer integration and both single- and double-reading workflows. It also states that the product is not available for sale or distribution in every country.

INSIGHT CXR addresses chest X-rays. Its current description combines abnormality detection with a separate normal-case identification engine under the DualScan name. That is a workflow proposition as well as a model proposition: the software is intended to influence how readers organize attention across a large volume of studies.

SCOPE IO analyzes tumor microenvironment features from H&E pathology slides for biopharma research. Its public FAQ explicitly identifies it as Research Use Only. The ability to investigate biomarkers and treatment-response associations should therefore not be described as an authorized diagnostic test or an autonomous method for selecting a patient’s therapy.

Lunit’s November 2025 announcement says Volpara now operates under the Lunit brand following its acquisition. The current site includes screening quality and workflow products alongside INSIGHT and SCOPE. This is one company identity with a broader portfolio, rather than a reason to count the acquired brand as another independent AI company.

02 / AudienceRadiology departments and biopharma teams require different evaluations

A breast-imaging service may be interested in reading assistance, image quality and screening operations. Its questions concern which studies are supported, how information appears during interpretation and whether the software fits the local reading protocol. A pathology research group instead needs reproducible image analysis and a defensible relationship between measured features and the scientific hypothesis.

The current portfolio presents both cancer screening and precision oncology, with an international customer footprint. This breadth and the integrated Volpara business explain Lunit’s prominence in applied medical AI. It does not mean one purchase supplies every application, or that a clinical authorization for one product transfers to the research portfolio.

PathAI is useful related coverage for digital pathology and AI-supported image analysis. Owkin offers an adjacent view of AI in biomedical research and drug development. Those comparisons help a research buyer locate SCOPE’s role; they are less direct substitutes for a radiology department choosing a mammography reading aid.

The least suitable approach is to choose Lunit solely because a presentation promises earlier cancer detection. A team needs to identify the specific clinical or research job and the outcome it can actually measure. Screening recall behavior, reader agreement and biomarker reproducibility are different endpoints with different evidence requirements.

03 / WorkflowA proposed mammography evaluation preserves the existing reading protocol

For a proposed INSIGHT MMG evaluation, begin with the clinical service’s current reading workflow and the exact available product version. Obtain its instructions for use and confirm the supported image types, population and jurisdiction. The responsible clinical team should define the dataset and review method; this article is not a recommendation to change a screening protocol.

Choose an approved retrospective set that reflects the service’s ordinary images and relevant variation. Include cases where a prior examination is available and cases where it is not, if those are part of the intended workflow. Record the information that a reader would normally see so the evaluation does not accidentally compare different clinical contexts.

Inspect how AI information arrives in the image viewer. Can the reader distinguish the original image from an overlay, understand which region is being highlighted and review the output at the appropriate point in the local protocol? A technically available score has limited value if its meaning or timing is unclear during interpretation.

Have the team record where AI output changes attention and where it adds unnecessary review. A highlighted region is a prompt for professional assessment, not a final diagnosis. Equally, an image without a highlight should not be treated as a universal guarantee of normality. The practical goal is to understand how the aid interacts with the reader’s established responsibilities.

For a service also considering quality or workflow products from the former Volpara portfolio, evaluate those components separately. An image-quality intervention can affect the inputs to a reading aid, while a workflow dashboard may address staffing or operational trends. Combining their results into one undifferentiated AI benefit would obscure which change actually helped.

A SCOPE IO evaluation requires a different plan. A research team might define a pathology cohort, inspect slide quality, compare extracted features with a reference annotation process and assess repeatability across the permitted scanner and file types. That is a proposed research workflow, not evidence that SCOPE is authorized for routine diagnostic decisions.

Sequenced has not performed either evaluation. The purpose of these examples is to make the product distinction concrete: one route examines a professional imaging workflow, while the other examines a research measurement process. The organization should judge each using the evidence standard appropriate to its intended use.

04 / PricingCommercial scope follows the product, geography and integration

Lunit’s reviewed pages direct prospective buyers to speak with a specialist. They do not establish a universal public list price for the whole portfolio. A quote should identify the application, version, supported country and deployment arrangement, along with the integration and support included.

For a screening service, clarify whether the proposal includes only an INSIGHT application or also quality, analytics and patient-workflow products. The unified brand makes the portfolio easier to find, but it does not prove that every component is bundled. Compare quotations against the same operational need and supported study volume.

For research work, specify the analysis service or software, data rights, supported image formats and the deliverables required for the project. The SCOPE IO page positions the product around biopharma and biomarker development. A research collaboration can involve responsibilities that differ substantially from a clinical site licence.

Implementation effort belongs in the comparison. Viewer integration, staff training, data preparation and local review all affect the total work required. A broad claim of efficiency should be translated into the particular step the buyer wants to improve, and evaluated without assuming that every minute saved in image processing becomes additional clinical capacity.

05 / DistinctionsThe portfolio joins image analysis with the surrounding screening process

The Volpara integration gives Lunit a broader screening story than a standalone detector. Its brand announcement links breast health, AI detection and precision oncology under one identity. The practical opportunity is to consider image quality, interpretation support and workflow together while retaining separate measures for each.

The research side asks a different question: can features in routine pathology images support a useful scientific hypothesis? SCOPE IO’s product description discusses spatial and cellular analysis of the tumor microenvironment. The valuable output is an inspectable measurement linked to a research question, not merely a visually impressive overlay.

Lunit publishes numerous clinical and technical performance claims. This blueprint does not convert those claims into a universal accuracy score. A buyer should inspect the original study, the software version, the comparison and the population. Evidence for one image modality or application cannot establish the performance of the entire company portfolio.

06 / QuestionsResearch-use labels and country restrictions must travel with the product

Is the exact application available in the proposed country? Both reviewed INSIGHT product pages warn that availability is not universal. A broad international footprint does not answer that question. The organization should obtain current product-specific labeling and confirm which functions are included in the version offered locally.

Is a SCOPE output intended for research or a diagnostic procedure? The SCOPE IO FAQ is explicit about Research Use Only, and its related-product cards carry similar restrictions. Marketing language about companion-diagnostic development describes a development context; it should not be read as blanket authorization for current clinical decision-making.

Which documentation applies to the former Volpara products? Lunit’s regulatory information page identifies Lunit International Limited as manufacturer for specified Volpara-branded products and describes how customers obtain instructions. That page is scoped to those products, so it should not be used as a universal regulatory statement for every INSIGHT or SCOPE application.

How will changes be assessed? The buyer should establish how model updates, acquisition changes and workflow adjustments are communicated and reviewed. For research, preserve the analysis version and preprocessing choices so results can be interpreted later. For clinical use, follow the institution’s process for evaluating a changed device within its authorized scope.

07 / DecisionChoose the application whose evidence matches the job

Lunit is a relevant candidate for screening services seeking imaging assistance and for biopharma teams evaluating pathology AI research. Those routes share a company but require distinct product choices, contracts and evaluation plans. Begin with the exact job and use the portfolio to identify the appropriate component.

A useful decision rests on current availability, inspectable outputs and evidence matched to the local purpose. Keep research promises, clinical functionality and operational improvements separate until each can be demonstrated in its own right.