Lumea

The lowest centrality in the pathology lane, and the company's own founding argument explains why.

The founding insight was that tissue handling has to be fixed before software can help, and the product line follows it literally. A sectionable matrix holding six biopsy cores, an orientation preserving board and frame, radio frequency identification tags for chain of custody, embedding tampers, an anti slip platform: these are physical devices, protected by six cited patents, and the headline benefit is a 14.5 percent average increase in histologic tissue surface area. No model is involved anywhere in that, and it is the layer the company describes as foundational.

The cleared software is an image management and viewing system. Viewer+ earned its 510(k) on rendering multi gigabyte images without latency and displaying them adequately for primary diagnosis, which is demanding engineering and is not machine learning. BxLink is a laboratory information system handling billing, specimen tracking and test ordering.

Two model claims exist and both are narrow. BxCamera applies artificial intelligence at the grossing station to standardise measurement and description and document specimens automatically, which is real image and document work in the laboratory rather than in diagnosis. And the diagnostic models are other companies' entirely: the marketplace integrates partners including Paige and Valar Labs, both separately indexed here.

Graded D. A pathologist using this platform to reach an artificial intelligence assisted diagnosis is using somebody else's model delivered through Lumea's screen, and the company's own contribution is the tissue, the workflow and the display.

The automation here is placed almost entirely in physical and clerical work, and the diagnosis is left alone deliberately.

Nothing in the first party product line makes a clinical determination. The tissue technology orients and holds specimens. Radio frequency identification tracks them. BxCamera standardises measurement and documentation at the grossing station, which replaces manual transcription rather than any judgement. The viewer displays images. The information system moves cases and orders. A pathologist reads the slide and signs the case, and the platform's contribution is that the tissue arrives properly oriented and the image renders without latency.

That placement is the strength. The company's argument is that variability enters at specimen handling and transcription rather than at interpretation, and it has automated exactly those steps while leaving interpretation to the pathologist. Chain of custody tracking adds an audit trail that makes the automated steps reviewable after the fact, which is oversight in the practical sense.

The clearance reinforces it. A viewer cleared for primary diagnosis has been assessed on whether pathologists reading through it reach equivalent conclusions to reading glass, which is a regulator confirming that the human remains the decision maker and the system does not degrade their judgement.

What holds this at B is the marketplace. Partner algorithms delivered into the workspace do make clinical assessments, and their oversight behaviour, confidence handling and failure modes belong to those partners. Nothing describes what the platform requires of an algorithm before integrating it, or how a result is presented to a pathologist so its provenance and uncertainty are visible.

Graded B.

Physical and engineering transparency is excellent and specific, model transparency is minimal, and the balance still earns a good grade because most of this product is not models.

On the physical side the disclosure is exact. Six United States patent numbers are cited for the tissue technology, the matrix is described as sectionable and holding up to six biopsy cores through processing, orientation preservation into the paraffin block and onto the slide is explained mechanically, and a specific compatibility statement is published: the matrix material does not interfere with immunohistochemistry, in situ hybridisation, fluorescence in situ hybridisation, sequencing or other molecular methods. That last statement is exactly what a laboratory director needs and would otherwise have to test, and publishing it plainly is good practice.

On the software the transparency is regulatory rather than technical. A published clearance number lets anyone retrieve the summary and read what was assessed, which is a stronger form of disclosure than a marketing description. The company also identifies the technical criterion that matters for viewer adoption, streaming multi gigabyte images without tiling or stuttering, rather than describing performance in adjectives.

On models there is almost nothing. BxCamera is described as artificial intelligence driven with no architecture, training data, accuracy figure, versioning or update cadence, and the marketplace algorithms are described by partner name rather than by capability detail.

Graded B: unusually concrete about the parts that are physical and regulated, thin about the small part that is machine learning.

Diagnostic model provenance is fully traceable because the models are openly other people's, and the physical supply chain is documented through patents.

The marketplace is the disclosure. Partner algorithms are named rather than white labelled, with Paige and Valar Labs both announced publicly and both separately indexed here, so a laboratory can follow any hosted model to its developer's own record and assess it independently. An integration with a diagnostics partner for a combined laboratory information and digital pathology offering is similarly named. A vendor that delivers third party intelligence under those companies' names, rather than presenting it as its own capability, has given customers everything they need to trace what they are running.

The physical supply chain is documented to an unusual standard. Six United States patent numbers are cited for the tissue technology, patents are public and examined, and the compatibility statement that the matrix material does not interfere with immunohistochemistry, in situ hybridisation or sequencing is a materials disclosure a laboratory can rely on and would otherwise have to validate itself.

What is missing keeps this from an A. BxCamera is described as artificial intelligence driven with no statement of whether the model is built in house or licensed and no training data description. And no contractual terms are published governing what marketplace partners may retain from images their models process, which is the same gap identified on the neighbouring platform record.

Graded B.

The strongest evidence base in the pathology lane, and unusually it spans regulatory, peer reviewed and operational sources rather than resting on one.

The regulatory evidence is the anchor. Viewer+ holds FDA 510(k) clearance for primary clinical diagnosis, granted March 2025 under number K242244, and the company publishes the number so a reader can verify it in the public database. Clearance for primary diagnosis on whole slide images requires a demonstrated equivalence study against glass, conducted to a defined protocol with pathologist readers, so a substantial clinical validation exists in the regulatory record whether or not it appears in a journal.

The peer reviewed evidence sits on the tissue technology, with published work cited on efficiency improvements from the biopsy chip and a specific quantified outcome, a 14.5 percent average increase in histologic tissue surface area. Surface area is a directly measurable physical quantity rather than a workflow estimate, which makes it a harder number than most in this index.

The company also engages honestly with the wider literature, stating that multiple peer reviewed studies show diagnostic concordance between digital and glass is equivalent and in some applications improved with artificial intelligence assistance, and adding that accuracy in both digital and traditional pathology depends significantly on specimen quality. That last sentence is self serving and it is also true.

Customer evidence is named and attributed to identified pathologists, and the specialty groups described are urology, gastroenterology and dermatology.

Graded A: a cleared indication requiring clinical validation, published quantitative results on the physical products, and honest positioning against the general literature.

Nothing published addresses retention, secondary use or model training, and the position this platform occupies makes it a more consequential silence than the modest first party artificial intelligence would suggest.

The platform sits at the confluence of everything: images from the scanner, the full laboratory record from the information system, physical specimen tracking from collection, molecular results ordered from outside laboratories, and outputs from partner algorithms. A laboratory running BxLink and Viewer+ has assembled, inside one vendor's environment, a linked dataset of tissue images and molecular results tied to diagnoses and outcomes. That linkage is the most commercially valuable asset in computational pathology, because image plus molecular plus outcome is exactly what model development requires and exactly what is hardest to assemble.

Nothing states whether the vendor retains any interest in it, whether de identified data may be used for product development, whether partner algorithm vendors may retain anything from images they process, or what happens to the archive at contract end.

The partner marketplace makes the third party question concrete rather than hypothetical. When a customer adopts an integrated algorithm, patient images are processed by a company the customer may not have contracted with directly, and only the platform could describe the terms. It does not.

One architectural point works in the vendor's favour and deserves recording. Chain of custody tracking from the operating room onward is a stewardship mechanism in the literal sense, providing a documented record of where a patient's tissue has been at every step, and that is better provenance than most laboratories have.

Graded D on the absence.

No published position was located, and the breadth of what this platform touches makes the gap wider than the usual software case.

BxLink is a full laboratory information system. It manages billing, specimen tracking, test ordering and reporting, which means it holds patient demographics, insurance and payment information, specimen and diagnostic data and the final pathology report. Viewer+ holds whole slide images with case and patient context. Radio frequency identification links physical tissue to a patient encounter from the moment of collection. Molecular test ordering sends patient material and orders to outside laboratories. Partner algorithms process patient images inside the workspace. That is a wider protected data estate than any other pathology record in this index carries.

Nothing published addresses it. No business associate agreement is offered or described, no protected data handling summary exists, no retention position is stated, and nothing describes what binds the marketplace partners whose models and tests touch patient material.

The molecular ordering path deserves specific mention because it is the one that crosses an organisational boundary by design. A one click order sends a patient's tissue and identifiers to a third party laboratory, and nothing describes the data flow, the consent basis or the agreements involved.

One indirect indicator exists and is not disclosure. A cleared diagnostic device sold into United States clinical laboratories operates within an accreditation regime that requires protected data controls, so the customer base has evidently satisfied itself. The next buyer has nothing published to work from.

Graded D.

No published security posture was located. No trust centre, no service organisation control claim, no HITRUST claim, no independent audit reference, no penetration testing statement, no vulnerability disclosure policy and no incident notification commitment were found. A privacy policy exists in the site footer and nothing else of this kind.

The gap is larger here than the absence alone suggests, for three reasons specific to this vendor.

The cleared viewer is a regulated medical device used for primary diagnosis, and premarket cybersecurity documentation has been expected for devices of this class for several years, so material almost certainly exists in the submission. None of it is public.

The laboratory information system holds billing and insurance information alongside clinical data, which is the combination attackers target, and information systems of that kind are among the most disruptive things a laboratory can lose.

And the partner marketplace means outside algorithms and molecular ordering integrations touch patient data inside the environment, with no described control boundary around them.

The contrast within the lane is worth recording plainly. This vendor holds the strongest regulatory credential in the pathology category and publishes the least security assurance of the three workflow records built in this session, where one neighbour names service organisation control 2 alongside a diagnostic marking and the other publishes four credentials at a fraction of the size.

Graded D on published evidence rather than on any judgement about the underlying engineering.

The strongest and most carefully stated regulatory position in the pathology lane.

Viewer+ holds FDA 510(k) clearance for primary clinical diagnosis, granted March 2025, and the company publishes the clearance number so anyone can verify it in the public premarket notification database. Clearance for primary diagnosis on whole slide images is a demanding indication requiring demonstrated equivalence to conventional microscopy, and it is the difference between a laboratory being able to sign out cases digitally and merely being able to view images. Several records in this lane are research use only or rely on a European marking, and one signals its United States status with an asterisk.

What lifts this to an A is the discipline around the edges of the claim rather than the claim itself. The company states on every page, in the site footer rather than in fine print behind a link, that BxLink has not been approved by the FDA for digital primary diagnosis and refers readers to the professional body's guidelines on digital pathology use. So the vendor holding the strongest clearance in the category is also the one most explicit about which of its products the clearance does not cover. That is the correct behaviour and it is rare: the commercially convenient move would be to let the cleared status colour the whole catalogue.

The company also publishes accurate context on the regulatory pathway, describing the 2017 de novo clearance of the first such system and the subsequent enforcement policy expansion, which is a fair account rather than a marketing one.

Graded A.

Nothing was located. No model card, no training data description, no accuracy figures for BxCamera, no subgroup analysis and no bias statement.

The first party exposure is modest because the first party models are modest. BxCamera standardises measurement and documentation at grossing, and if its performance varies with specimen type, size or staining the consequence is inconsistent documentation rather than a wrong diagnosis.

The more interesting exposure is structural and runs through the physical products, which is unusual enough to record. The tissue technology is engineered around specific specimen types, with the biopsy chip built for six prostate cores and separate products for bladder, gastrointestinal and cytology. Physical products designed around a protocol work best when the protocol is followed, and biopsy practice varies between operators and institutions. Nothing published examines whether the stated benefits hold across that variation, and a device that improves tissue surface area by 14.5 percent on average says nothing about the distribution behind the average.

The marketplace carries the conventional version. Partner algorithms are validated by their developers on their own cohorts, and a laboratory adopting one through this platform inherits whatever population limitations it carries. Unlike Gestalt on the adjacent record, no evaluation tool is offered that would let a customer test a hosted algorithm on their own material.

Graded D on the absence, with the note that the harm profile here is lower than for a diagnostic model vendor.

Nothing published addresses liability directly, and the regulatory posture does more work here than a contract clause would on most records, which is why this grades above the session floor.

The clearance establishes an allocation rather than merely a credential. A viewer cleared for primary clinical diagnosis has been assessed as adequate for a pathologist to reach a diagnosis through, and the pathologist signing the case remains the responsible clinician. That is a defined and understood division: the device performs to its cleared indication and the physician exercises judgement. The company reinforces it by stating plainly that BxLink is not approved for primary diagnosis, so a laboratory that uses the uncleared product for diagnosis has departed from the stated indication and knows it.

The physical products carry a different and more concrete exposure that is unaddressed. A tissue matrix, board or frame that fails during processing damages or disorients a specimen, and a biopsy specimen is not repeatable. Nothing describes product warranty, defect handling or recourse for a compromised specimen, and this is the vendor whose founding argument is that specimen integrity determines diagnostic accuracy.

The chain of custody tracking is a genuine mitigation and deserves credit. When tissue is tracked from the operating room onward, a laboratory investigating a lost or mislabelled specimen has a record rather than a reconstruction.

The marketplace division is unaddressed as elsewhere: when a partner algorithm informs a diagnosis, nothing states what the platform warrants about it.

Graded C.

Two deployment paths are offered and the company is honest about which suits whom, with the ceiling set by an absence of named systems.

The architecture accommodates both directions a laboratory can take. Viewer+ integrates into an existing laboratory information system without replacing it, so a laboratory that has invested in its current system keeps it. BxLink replaces the system entirely for laboratories without one or ready to change, combining viewer and information system so image and case workflows stop running in parallel. The company states the choice rule explicitly rather than steering every prospect toward the larger sale, which is both commercially honest and genuinely useful.

Interoperability extends beyond the record in directions most of this lane does not reach. Molecular test ordering sits one click from the diagnostic workspace, connecting the pathologist to outside reference laboratories. The partner marketplace spans algorithms, molecular tests and scanners, so scanner neutrality is implied. Radio frequency identification links physical specimens to digital records, which is interoperability between the bench and the database rather than between two databases.

What holds this at B is specification. No laboratory information system is named for the integration path, no record system is named, no interoperability standard is cited, and no application programming interface was located. Against Gestalt on the adjacent record, which names Epic and Cerner and cites a completed migration between proprietary systems, this is asserted rather than evidenced.

Graded B.

Nothing published on hosting, region or residency, and this record has a deployment dimension no software vendor in the index shares.

That dimension is physical. The tissue technology deploys into operating rooms and grossing stations as consumable devices, boards, frames, chips, tampers and radio frequency identification tags, so implementation involves changing how surgeons load specimens and how histotechnologists embed them. A distributor network serves Europe and international markets, and product catalogues are issued regionally, which is a supply chain rather than a software rollout. None of that raises residency questions and all of it raises deployment ones the company describes only in product terms.

On the software nothing is stated. No hosting provider, no region, no residency option, no subprocessor list, no retention position and no export terms were located for either the viewer or the laboratory information system. The viewer is described as giving pathologists full diagnostic capability on any device from any location, which implies a hosted or streamed architecture, and the rendering claim of milliseconds for multi gigabyte images implies substantial infrastructure, and neither is described.

Two consequences deserve naming. A laboratory information system is a system of record with retention obligations measured in years and sometimes decades, and nothing describes data export or what happens at contract end. And a viewer cleared for primary diagnosis becomes part of the diagnostic pathway, so availability is a clinical dependency rather than a convenience, and no uptime commitment, status record or continuity position was located.

Graded D on the absence.

The best commercial disclosure in this session, and the reason is simple: this vendor operates an online shop.

A purchasing channel is published for North America, with product catalogues issued for North America and Europe, an international distributor network, and a partner marketplace covering artificial intelligence, molecular tests, scanners and laboratory technology. A laboratory can see the consumable and device range and buy through a published channel rather than beginning with a sales conversation. Nothing else in the pathology lane offers that, and very little in this index does.

The product architecture is also explained in commercially useful terms rather than marketing ones. The company states the decision rule directly: if a laboratory has no existing information system or wants to replace one, BxLink is the answer, and if it has a system it intends to keep, Viewer+ integrates into it without replacement. Telling a prospect which of your two products not to buy is unusual and it is the sort of clarity that shortens an evaluation.

What is missing keeps this from an A. No figures were retrievable for the software platforms, no subscription or licence structure is described for Viewer+ or BxLink, no implementation cost is stated, and the marketplace does not indicate whether partner algorithms and molecular tests are licensed through Lumea or separately. The published shop appears to cover consumables and devices rather than the platforms, so the software side remains a quote.

Graded B on the strength of the published purchasing channel and the honest product selection guidance.

Deliberately narrow by specialty, complete within those specialties, and covering more of the diagnostic path than anything else in this index.

The specialties are named and the products are built per specialty rather than configured for it: prostate, bladder, gastrointestinal and cytology tissue technology, with solutions material for pathology, dermatology, gastroenterology and urology. Prostate is the anchor and the biopsy chip is designed around six core specimens, which is the prostate biopsy protocol. That is genuine specialty engineering rather than a generic platform with presets, and it is why the customer base is described as urology, gastrointestinal and dermatology groups.

What is unusually broad is not specialty but stage. The platform covers the operating room where tissue is loaded onto a board, the grossing station, laboratory processing, the scanner, the diagnostic workspace, molecular test ordering and the final report. No other record in this index spans from specimen collection to report, and the chain of custody tracking that runs through it is a coherent consequence of covering the whole path.

Institutional coverage is mid market. Individual pathologists, practices and specialty groups are named, with a stated aim of scaling for individual pathologists and practices, and a distributor network reaching Europe and international markets.

What holds this at B is the absence of the general hospital laboratory. This is a specialty group platform, and a large academic department with a broad case mix across every organ system is not the buyer this is designed for.

Graded B.