Artrya

The algorithm is the product and the company built the business around a single technical bet. Salix is a CCTA image analysis platform where AI identifies and quantifies plaque type and extent, with plaque volume computed by identifying voxels within target Hounsfield Unit ranges spatially located between the lumen and outer vessel wall. The company states the algorithm was built from thousands of scans sourced from medical institutes in the US, Australia and Canada.

No hardware line and no services business; notably it uses no external reading teams to validate scan data, so there is no human processing layer beneath the AI. The plaque module reportedly accounts for around 70 percent of projected group revenue.

A on a structural design choice the company states explicitly: Salix uses no external reading teams to validate cardiac scan data, so physicians retain direct control of AI supported scans and results return in under 10 minutes.

That is the opposite architecture to the analysis service model, where scans leave the institution for vendor side processing and return hours later. Removing the human processing layer makes the autonomy claim honest and auditable rather than obscured, and it is the reason the point of care timing claim is credible.

Use as the contrast against Heartflow and Elucid, both of which involve vendor side processing steps whose extent is undisclosed.

More mechanically specific than most imaging vendors. The plaque volume computation is described concretely as identifying voxels within target Hounsfield Unit ranges spatially located between the inner lumen and outer vessel wall, and the reported plaque features are enumerated by name including low attenuation plaque, spotty calcification, positive remodeling and napkin-ring sign.

Training data provenance is stated at thousands of scans from institutions across the US, Australia and Canada. Graded B rather than A because the proprietary Plaque Dispersion Score, which is the differentiating metric and the subject of the SAPPHIRE study, has no published derivation or validation, and no accuracy figures with stated methodology were located.

An unusual posture rather than an empty one, and the ambiguity is what holds the grade. Three product specific commitments are stated and they are stronger than most in this category: users are contractually told not to upload personal information about any person except as agreed in the customer agreement, anything uploaded will be used only to administer that user's use, which on its face excludes research and model development, and an account holder can request erasure of all associated data through a named officer, with records no longer required deleted or permanently de identified.

A use limitation that excludes model development is exactly what this axis asks for. What is missing is scope and duration, and the first commitment undercuts the second in an odd way: because the default instruction is not to put personal information into the product at all, it is unclear which document actually governs an identified cardiac study, which is what the product exists to analyse.

A term protecting data that the user was told not to upload may or may not reach the data they necessarily did. The same policy separately permits disclosure to a third party who may combine data to develop anonymised consumer insights, and lists advertising and marketing among its purposes, with no clinical carve out.

Establish which agreement governs patient imaging, whether studies are de identified before or after upload, how long they persist, and that the consumer insights permission cannot reach clinical material.

The weakest axis and the honest gap in an otherwise strong record. No peer reviewed validation study for Salix was located. The SAPPHIRE study is the evidence programme, but it is prospective in the sense of not yet reporting: announced in 2025, expected to commence early in calendar 2026, structured as a retrospective multi-centre real-world study in three phases across six US health systems including Piedmont Healthcare.

Broker commentary refers to an earlier Australian study with positive results that SAPPHIRE expands upon, but that study was not located in published form. So the company has a serious evidence plan and named academic partners, and nothing published to grade yet. Contrast Heartflow with 600+ publications and Caristo with a Lancet cohort, both indexed in the same lane. Worth revisiting once SAPPHIRE reports.

An unusual posture rather than an empty one, and the ambiguity is what holds the grade. Three product specific commitments are stated and they are stronger than most in this category: users are contractually told not to upload personal information about any person into the software except as agreed in the customer agreement; anything that is uploaded will be used only to administer that user's use, which on its face excludes research and model development; and an account holder can request erasure of all data associated with their account through the Privacy Officer.

Records no longer required are deleted or permanently de identified. What is missing is scope and duration. No retention period is given, and because the default instruction is not to put personal information into the product at all, it is unclear which document actually governs an identified coronary CT study.

The same policy separately permits disclosure to a third party who may combine data to develop anonymised consumer insights, and lists advertising and marketing among its purposes, with no clinical carve out. Worth establishing which agreement governs patient imaging, whether studies are de identified before upload or after, and how long they persist.

The governing regime named is the Australian Privacy Act 1988 and the Australian Privacy Principles, with a Privacy Officer as the contact point. HIPAA and the business associate agreement are absent entirely. The gap is pointed rather than theoretical: the US product is FDA cleared under K243038 and the company publishes its expected US reimbursement pathway, so American clinical deployment is central to the commercial plan while the only published privacy instrument is bound to Australian law.

Australian Privacy Principles are a genuine regime and this is not an absence of governance, but they are not the regime a US health system is surveyed against. A US buyer should establish HIPAA posture and BAA availability directly, as nothing published speaks to either.

No SOC 2, ISO 27001, HITRUST, trust center or penetration testing statement was located. The privacy policy describes physical, administrative, personnel and technical measures in general terms, with two specifics worth crediting: servers holding personal information are stated to be encrypted, and records no longer required are deleted or permanently de identified rather than retained by default.

Against that, the same section disclaims security of transmission and states that information transmitted to the company is sent at the user's own risk, without distinguishing website contact data from clinical imaging. For a product that receives coronary CT studies that distinction matters. Nothing here has been externally tested, so a buyer is relying on description rather than attestation. Worth asking whether any independent security audit exists and what its scope was.

Clean, well documented and disciplined about scope, which is what earns the A despite the company being much smaller than its lane competitors. Salix Coronary Anatomy cleared March 2025; Salix Coronary Plaque cleared August 2025 with the 510(k) number publicly identified as K251837 and the FDA summary document available, which is more specific than most vendors in this index provide; Salix Coronary Flow explicitly described as completed and undergoing clinical data validation ahead of a future 510(k) submission, not as available.

Naming which module holds which clearance and stating plainly that the third is not yet submitted is the same disclosure discipline that earned Milvue credit. Note for completeness that a Nature Reviews Cardiology benchmarking publication listed Artrya Salix Coronary Plaque as Research Use Only version 1.0 at the time of that analysis, which predates the August 2025 clearance.

Better than the imaging lane norm for a specific and creditable reason: the SAPPHIRE study is designed with an explicit focus on coronary artery disease in WOMEN, and the company states the rationale, that a large majority of women who die from coronary artery disease showed no prior symptoms. Sex differences in coronary disease presentation are among the best documented disparities in cardiology and are routinely ignored by imaging AI vendors.

Designing the flagship evidence programme around that gap is a substantive governance choice rather than a policy statement. Graded B rather than A because no subgroup performance data exists yet, the study has not reported, and no analysis by race, ethnicity or body habitus was located. Compare Cognivue's FOCUS study, which is the benchmark because it reported.

The standard measurements are described with real mechanical specificity and the differentiating one is not, which is the pattern to note. Plaque volume computation is described concretely as identifying voxels within target attenuation ranges spatially located between the inner lumen and the outer vessel wall, the reported plaque features are enumerated by name rather than summarised, and training data provenance is stated as thousands of scans from institutions across three countries, which lets a reader judge geographic breadth.

All of that concerns measurements that are already established in the literature and that a competitor could reproduce. The proprietary dispersion score, which is the metric that differentiates the product and the subject of its named study, has no published derivation or validation.

That inversion is worth naming because it recurs: a vendor discloses generously about the parts anyone could replicate and withholds the part a buyer is actually paying for, so the disclosure that exists cannot substitute for the one that does not. A score with no published derivation cannot be interrogated when it disagrees with a cardiologist's reading, and a clinician deciding against it has nothing to point to.

No accuracy figures with stated methodology were located and no warranty, indemnity or remediation commitment attaches. Ask for the derivation of the proprietary score, its validation, and what a clinician should do when it conflicts with conventional assessment.

Modules share a single user interface, so the plaque capability was enabled inside the already-deployed anatomy platform immediately on clearance with no separate integration project, which is a real deployment advantage of the modular design. Full integration reported across five Tanner Health hospitals. Graded B rather than A because no named PACS, RIS or EHR integrations were disclosed and the interoperability claim rests on internal platform coherence rather than documented third party interfaces.

Cloud platform delivering point-of-care results in under 10 minutes without requiring clinicians to change or use multiple systems. The absence of external reading teams means no queue behind the analysis, which is what makes near real-time credible.

Graded B rather than A because it is cloud-only with no on-premise option located, patient imaging necessarily leaves the institution, and no data residency terms were published despite the company operating across Australian and US jurisdictions.

The most legible commercial model in the cardiovascular CT lane and among the clearest in the index, largely because ASX listing forces disclosure. The company states its pricing basis directly as a fee per scan assessed, reports that scan volumes at Tanner Health are generating fee-per-scan revenue, and quantifies the reimbursement environment precisely: the existing CPT code for automated plaque analysis from CCTA was upgraded to Category I in October 2024 at US$950 per assessment, taking full effect 1 January 2026.

It also discloses the commercial dependency plainly, that the plaque module represents roughly 70 percent of projected group revenue. A buyer can determine the pricing mechanism, the reimbursement rate and the vendor's own commercial exposure from public sources, which is rare anywhere in this index.

Narrow and deliberately so: coronary artery disease assessment from CCTA, serving cardiology and cardiac imaging, currently through a small number of named US hospital systems plus Australian operations. Two cleared modules with a third pending. Graded on the same scale as index vendors spanning multiple body systems, and the commercial footprint is materially smaller than Heartflow, Cleerly or Elucid in the same lane.