Elucid

The product is the algorithm. PlaqueIQ has no meaningful non-AI version: the entire commercial claim rests on convolutional neural networks classifying plaque components, and the company describes itself as an AI medical technology company rather than an imaging company that added AI. No underlying services business or hardware line dilutes this.

A human step in every case, a review and rework path, and no published limits on where the analysis should not be used. Clinicians send studies to Elucid with one click; Elucid analysts generate the segmentation as a service, image restoration reduces motion and calcium blooming, and the models classify and quantify plaque.

Physicians review results layered over the raw CT in an interactive viewer and a physician-validated PDF report; the 510(k) summary says they cannot edit the analysis but can request rework, and that Elucid monitors analyst performance for repeatability. The clearance states the results support clinicians alongside history, symptoms and other tests. The company does not say in which patients, scans or anatomy the analysis was not evaluated, or how often analyses are reworked.

The method and its ground truth are described, and versions are visible; training data size and update practice are not published. Elucid says Plaque-IQ was trained and validated on ground-truth histology, classifying plaque components from the spatial distribution of tissue properties across neighboring voxels rather than fixed Hounsfield unit ranges, because those ranges overlap between lipid rich necrotic core, intraplaque hemorrhage and other tissue; published work describes convolutional networks trained on carotid endarterectomy specimens co-registered to CT angiography.

The regulatory record shows the product line's versions (vascuCAP in 2017 and 2018, ElucidVivo A.3 in 2022, PlaqueIQ B.1P in 2024), and image restoration algorithms are applied to reduce motion and calcium blooming. There is no count of the specimens and patients the models were trained on, no account of where they came from, and no description of how model changes are validated and communicated.

A short chain for the models, with people in it and the host unnamed. Elucid builds its own plaque models, cleared in its own name, so no outside model provider sits between a CT study and a result. The chain includes Elucid analysts, who segment every study in the cloud service before the models run.

The privacy materials say onward transfers to agents carry contractual assurances of equivalent protection and that some products may need remote access from the US to data of patients in the European Economic Area. Neither the cloud host nor the agents are named, there is no subprocessor list, and nothing says whether customer studies are used to develop algorithms, which the privacy terms permit.

Validation against tissue is published; evidence that use changes outcomes is not. Elucid publishes correlations between Plaque-IQ and pathologist-read histology from the data behind its clearance (K241524): total plaque r = 0.99, calcified 0.95, non-calcified 0.86, and lipid rich necrotic core 0.88, without the number of specimens or patients. Validation against removed tissue is a strong reference standard compared with agreement between readers.

The company lists recent conference abstracts, and Elucid appears among the named vendors in 2026 reviews in Radiology: Cardiothoracic Imaging, which also say reproducibility, added prognostic value and clinical impact remain open for the whole category. No prospective study shows that using Plaque-IQ changes treatment or outcomes. Published research short of independent validation of clinical benefit.

A dedicated section on protected health information sets out real measures: processing only as needed and under the direction of the customer, physical, administrative and technical safeguards, confidentiality training that applicable employees sign for, encrypted TLS transfer, a firewall, and access limited to authorized personnel with unique credentials. Onward transfers to agents carry contractual assurances of equivalent protection.

The open question is the one that matters most, and the company's own wording creates it: stated purposes include improving software, algorithms and features based on the data, information and feedback received, and conducting research and development using it, with nothing excluding patient imaging data from that use and no de identification qualifier attached. There is also no retention period for submitted studies and no deletion commitment on termination. Whether patient scans actually feed algorithm development, and on what basis, is not stated.

A privacy document reaches the service, and business associate status is not stated. Elucid's privacy materials include a section on protected health information describing processing only as needed and under the customer's direction, with physical, administrative and technical safeguards, and a Data Privacy Framework program that describes Elucid as a processor acting for a customer.

Neither HIPAA nor a business associate agreement is mentioned, although US health systems send identifiable CT studies to Elucid's cloud and its analysts work on them. A substantive document reaching the service with no statement of business associate status.

Nothing independent backs the stated controls. Elucid's privacy materials describe encryption at rest and in transit, vulnerability management, backups, a firewall, access limited to authorized personnel with unique credentials, and signed confidentiality training. No SOC 2 report, HITRUST or ISO 27001 certification, independent penetration testing or trust portal is published, and the site footer carries only privacy and cookie policies. Elucid is certified to the EU-U.S.

Data Privacy Framework, which is an FTC-enforceable privacy commitment with independent dispute resolution, not an audit of security controls. The service is cloud based and Elucid staff segment every study, yet nothing on the site says whether a SOC 2 or HITRUST assessment exists, what it covers, or how analyst access to studies is controlled.

The cleared product is identifiable and the pending ones are labeled as pending. Elucid PlaqueIQ was cleared on 12 September 2024 (K241524; 21 CFR 892.2050, medical image management and processing system, Class II, product code LLZ) to support qualified clinicians in non-invasive evaluation of plaque quantification and morphology from previously acquired CT angiography, used with the clinical history, other tests and clinical judgment.

It builds on earlier Elucid clearances (vascuCAP, K163071 in 2017 and K183012 in 2018; ElucidVivo A.3, K221463 in 2022). The company marks FFR-CT and stenosis as pending FDA review on its pages; its dashboard description refers to investigational FFR analysis derived from plaque, so that analysis is not cleared either. The carotid application is marketed under the same plaque analysis; the clearance's indication covers CT angiography without naming a vessel bed.

No demographic subgroup analysis is published. The gap carries extra weight because the ground truth training set is carotid endarterectomy specimens, meaning tissue from patients who underwent surgery, which is a specific and nonrepresentative population. Whether plaque classification performance holds across sex, age and ethnicity is not addressed in the published material, and cardiovascular risk stratification has well documented disparities.

Mechanisms let a physician check and query the output; nothing stands behind it and no error rate has a denominator. Elucid analysts segment each study, the physician reviews the result layered over the raw CT, and although clinicians cannot edit the analysis they can request rework; the 510(k) summary (K241524) says Elucid monitors analyst and product performance for repeatability and reproducibility.

The clearance summary reports only non-clinical verification and validation, and the histology correlations the company publishes (total plaque r = 0.99, lipid rich necrotic core 0.88) give no specimen or patient count, so a buyer has no error rate with a denominator to hold the product to. No warranty, indemnity or remediation commitment is published. Mechanisms to challenge the output, nothing behind it.

Results reach the physician in two stated ways: a DICOM wrapped PDF delivered to the customer's PACS, and a link to an interactive browser for interrogating the coronary tree and the segmentation in more detail. Wrapping the report as DICOM is a well chosen mechanism, because results appear inside the existing radiology workflow without requiring a new system for the reader. Beyond that the picture is thin.

No EHR integration is named, there is no mention of FHIR or HL7 result routing, no PACS vendors or integration partners are listed, and nothing describes how the analysis reaches the ordering clinician or the patient record. The integration work a site takes on is not described either.

The model is vendor side processing and the company states it plainly, including that certain products may require remote access from the United States to protected health information of customers' patients in the European Economic Area. That candor is useful: it answers where the scan goes and who can reach it.

What is absent is the detail underneath: no hosting provider or cloud region is named, no data residency commitment is offered to US customers, and there is no on premise or customer tenant deployment option. Certification to the EU-U.S. Data Privacy Framework supplies a lawful transfer mechanism for European data specifically. There is no word on where studies are stored, who can access them, or for how long.

No price is published; the billing codes and coverage are. Elucid publishes the CPT codes that apply (CCTA 75574; Plaque-IQ under Category III codes 0623T to 0626T and, from January 2026, Category I 75577) and a state by state insurer coverage map, and says 70 percent of insured US lives have positive coverage policies; four of seven Medicare contractors, EviCore and UnitedHealthcare cover AI plaque analysis.

Those describe what a provider can bill, not what Elucid charges: no per study price, license structure or volume terms are published, and the route in is a conversation. No price, with how the product is bought and paid for discoverable.

Two vessel beds, named, with tissue validation behind the plaque classification. Plaque-IQ analyzes coronary and, since October 2025, carotid CT angiography, reporting patient, vessel and lesion level plaque composition including lipid rich necrotic core, stenosis and CAD-RADS scoring for cardiologists, radiologists and vascular physicians assessing heart attack and stroke risk.

The plaque classification is validated against histology, and the company markets it for a broad insured US population. Which patient groups and scan conditions the analysis was validated in, and where it should not be used, are not published, and the FFR-CT and stenosis outputs are pending FDA review. The narrow scope is not the weakness; the missing validation by setting is.