Discharge, detect or page the team: three regulated jobs for imaging AI

Three products in Recent's library read a medical image and return something. DERM, from Skin Analytics, assesses photographs of skin lesions. qXR-Detect, from Qure.ai, reads chest X-rays. Viz LVO and Viz ICH, modules of the Viz Platform, analyse CT scans of the brain. All three have been assessed by a public body. The jobs they were assessed for are not the same. Under NICE guidance in England, a benign label from DERM can discharge a person from an urgent pathway. Under a US FDA clearance, qXR-Detect marks regions of an image for the clinician reading it. Under a different FDA regulation, Viz's modules notify a specialist and, by that regulation's definition, do not mark the image at all.

What follows sets out each job in the assessing body's words, then each company's description, then what the difference means for the interface. Recent has not tested any of these products, and nothing here is guidance on clinical use.

The three jobs, in NICE's and the FDA's words

DERM: a benign label that discharges

In NICE's description, DERM works within teledermatology services after referral from primary care, on images taken by a smartphone with a dermoscopic lens attachment, and can be used “as an automated tool or with a healthcare professional review (known as a second read)”. NICE states what its label does:

If DERM labels a lesion as pre-cancer or cancer, NICE says “an NHS dermatologist reviews the case virtually”.

The recommendation is conditional. DERM can be used “during the evidence generation period” only “if the evidence outlined in the evidence generation plan is being generated” and “once it has appropriate regulatory approval including NHS England's Digital Technology Assessment Criteria (DTAC) approval”. To mitigate “the potential risk of missed or delayed cancer diagnoses”, NICE's first measure is “doing a healthcare professional review for people with black or brown skin”. Its reason: “The evidence on DERM is mostly for skin lesions in people with white skin”, and “More evidence is needed to be certain that automated DERM does not incorrectly detect or miss skin cancer in people with black or brown skin.” NICE also asks for “regular monitoring of DERM's performance to maintain accuracy” and additional protocols when necessary, such as “a national governance framework to ensure local oversight of use of DERM”.

NICE gives the cost as “£30 per referral” and states that “DERM has a class III CE-mark certification”. The period lasts 3 years; at its end NICE “will review the evidence and assess if the technology can be routinely adopted in the NHS”, and “the recommendations may change”.

In June 2026 Skin Analytics launched DERM Zero. The company says it is CE marked to Class III and “delivers an autonomous clinical decision directly from a standard smartphone”, “without the need for a second review by a clinician”. Those are the company's statements. NICE assessed DERM, used with a dermoscopic lens inside teledermatology services; nothing in its guidance is about DERM Zero.

qXR-Detect: a mark for the reader

The FDA cleared qXR-Detect through 510(k) K251934 on 16 January 2026, under product code MYN, “Analyzer, Medical Image”. Its indications for use describe “a computer-assisted detection (CADe) software device that analyzes chest radiographs and highlights suspicious regions of interest”, sorted into six categories, for adults only. The summary adds that users “must not use the qXR-Detect generated output as the primary interpretation”. The regulation for this class of device adds that it “is not intended to replace the review by a qualified radiologist”.

Qure.ai's announcement uses the same term and gives the company's own count: “FDA cleared indications now total 26 across 9 products”. The same company holds clearances of another kind: openFDA lists qXR-PTX-PE (K230899) under product code QFM, “Radiological Computer-Assisted Prioritization Software For Lesions”, and the announcement lists “Triage and notification of pneumothorax and pleural effusion” among its existing clearances. The job belongs to the cleared device, not to the brand.

Viz: a notification that runs in parallel

Viz LVO and Viz ICH are cleared under product code QAS, “Radiological Computer-Assisted Triage And Notification Software”. Viz LVO's indications say it notifies a neurovascular specialist that a suspected large vessel occlusion “has been identified and recommends review of those images”. Images previewed on the phone “are compressed and are for informational purposes only”, and the device “should not be used in-lieu of full patient evaluation or relied upon to make or confirm diagnosis”. Viz ICH's indications use the same “notification-only” wording.

The Viz Platform is a bundle. openFDA also lists Viz records under product code QIH, “Automated Radiological Image Processing Software”, including K232363 (Viz HDS, Viz Volume Plus, Viz ICH+) and K250354 (Viz Subdural+), so modules on one platform carry different regulatory jobs. For the Viz Pulmonary Suite, launched in May 2026, the company says Viz PE “was shown to reduce time to treatment from 1.75 days to 0.56 days” in a single-centre study, and its own footnote says the results “may not be generalizable”. The release does not mention the FDA.

How each company describes the job

Qure.ai's product page describes the qXR suite as a whole and names no market; it shows an “FDA cleared” badge and a CE 0123 mark without saying which function each covers, so a reader cannot tell from it which function is cleared where. Skin Analytics' “Recommended” is a one-word summary of a conditional, time-limited recommendation. Viz's “triages” matches its regulatory category; “detects” is the company's verb.

Recent analysis: what the job does to the interface

This is Recent's reading of the sources above, not a finding of any regulator.

The jobs differ, first, in who receives the output. With DERM in the pathway NICE describes, a benign label can reach the person as a discharge, with the results and safety-netting advice, so the output is an outcome rather than a suggestion; NICE's conditions (a healthcare professional review for people with black or brown skin, performance monitoring, governance) sit around that outcome. With qXR-Detect, the mark goes to the clinician reading the X-ray, during the read, and may not stand in for it. With Viz LVO, the output goes to a specialist's phone as a notification with compressed images that are not for diagnosis, while the standard read continues regardless.

They differ, second, in what the output may look like. A highlighted region is what a detection device is for, and the triage regulation excludes it by definition. A designer borrowing a “benign” label, a highlighted region or a push alert from one of these products borrows a regulatory category with it. An earlier Journal article compared how companies describe human review; these three suggest that where the review sits follows from the job the product was assessed for.

NICE's guidance may change after its evidence period, and each FDA clearance covers the device and the indications for use named in it.