AI & Automation

Ultrasound-AI Explained: What This Clearance Changes

Jun 14, 2026

Ultrasound-AI is the on-cart automation of the two hardest parts of a routine scan — taking the measurements and classifying what is on the screen — so that the same exam comes out the same way regardless of which operator runs the probe.

That one-sentence definition is the thing to anchor on, because the term went from niche to cleared-headline in a single regulatory filing. On June 2, 2026, Philips announced FDA 510(k) clearance for its Elevate Plus software, which bakes automated measurement and AI lesion classification directly into the ultrasound cart. This page is the plain-English explanation of what ultrasound-AI is, what actually happened, why it happened now, and — separated cleanly into its own section — where we think it lands for small and mid-size operators over the next few years.

TL;DR

  • Philips received FDA 510(k) clearance for Elevate Plus on June 2, 2026, adding AI measurement and lesion classification on-cart, as detailed in the Philips press release.

  • Auto Measure hits over 93% accuracy versus manual expert measurement, according to the Philips press release, which reports up to 30% scan-time reduction.

  • The same software cuts scanning time by up to 30% on covered exams, per the Philips clearance.

  • The "why now" is workforce: sonographer demand is projected to grow much faster than average while the talent pool stays thin, so standardizing routine exams is an economic necessity, not a luxury.

  • This is a real, cleared product running on installed EPIQ Elite and Affiniti systems — not a demo. The honest limits are scope (only certain exams are automated), the human sign-off requirement, and the integration work behind the cart.

  • If you want the operator's view — daily tasks, costs, staffing — read the companion piece on what ultrasound-AI means for healthcare practices.

What actually happened

On June 2, 2026, Philips announced it had received FDA 510(k) clearance for Elevate Plus, a software upgrade for its EPIQ Elite and Affiniti general-imaging ultrasound systems. The Philips press release states the release also obtained CE Mark, meaning it is cleared on both sides of the Atlantic at once.

The headline capability is automated measurement. Philips' Auto Measure Abdomen feature performs routine abdominal measurements automatically, and according to the Philips press release, it reaches over 93% accuracy compared with manual measurements made by clinical experts. A clinical user at Boston Medical Center quoted in the same release said automating those measurement tasks lets sonographers "reduce scanning time by up to 30% without sacrificing clinical precision."

The second capability is decision support on the cart. Philips integrated Koios AI directly into the ultrasound system: the Philips press release reports Koios Bi-RADS returns a breast-lesion assessment in under 2 seconds, while Koios Ti-RADS leverages a library of more than 350,000 pathology-proven cases to classify thyroid nodules.

Timeline of the clearance

DateEventSourced detail
Dec 16, 2021Koios DS first FDA-clearedTrained on data from 48 global sites
June 2, 2026Elevate Plus announcedFDA 510(k) + CE Mark
As of June 2026Auto Measure accuracy93%+ vs. manual experts
As of June 2026Koios Ti-RADS library350,000+ pathology cases

The mechanism, in plain language

Strip away the jargon and a routine ultrasound exam is a series of repetitive, judgment-light steps wrapped around a few high-judgment ones. The sonographer finds the organ, freezes the right frame, drops calipers to measure it, labels it, and repeats. Then a physician reads the images and decides what they mean. The repetitive measuring is where time leaks and where two operators can produce two different numbers on the same patient.

Ultrasound-AI attacks exactly that gap. The Auto Measure software recognizes the anatomy in the frame and places the calipers itself, so the measurement does not depend on whose hand is on the probe. The Koios layer goes one step further: when a breast or thyroid lesion appears, it scores it against the radiology standards — Bi-RADS for breast, Ti-RADS for thyroid — and hands the clinician a structured risk read instead of a blank report field.

The reason this matters is that the bottleneck in imaging has never been knowledge — the scoring rules are published — it has been consistency and capacity. Koios DS, the engine inside this clearance, was originally built to align directly to the American College of Radiology's BI-RADS and TI-RADS rating systems and the American Thyroid Association's classification system, and according to AuntMinnie, it was developed using ultrasound data from a network of 48 sites worldwide. Moving that engine on-cart means the structured read happens during the exam, not hours later at a reading station.

Teams already routing imaging reports and referrals through US Tech Automations workflows will treat this as a model swap, not a rebuild — the structured Bi-RADS/Ti-RADS output drops into the same downstream routing they already run.

There is a subtle but important distinction in what "AI on the cart" means here. Earlier generations of imaging AI lived at the reading station: the sonographer scanned, the images traveled to a workstation, and software flagged findings there for the radiologist. Elevate Plus moves the assistance forward in time, into the exam itself. That is why the scan-time figure exists at all — the measurement is automated while the patient is still on the table, not reconstructed afterward. It also changes who interacts with the AI. Instead of a radiologist reviewing AI flags hours later, the sonographer sees the structured output during the study and can re-image immediately if something looks off, which is a meaningfully tighter feedback loop than the prior workflow allowed.

Why now — what constraint broke

The constraint that broke is staffing. According to the U.S. Bureau of Labor Statistics, employment of diagnostic medical sonographers is projected to grow 13% from 2024 to 2034 — much faster than average — with about 5,800 openings each year, while the same Auto Measure efficiency comes from the Philips clearance. When demand outruns supply, the only way to keep throughput up is to take routine work off the scarce specialist's plate.

The economics reinforce it. According to the U.S. Bureau of Labor Statistics, the median annual wage for diagnostic medical sonographers was $89,340 in May 2024, with the top 10% earning more than $120,960 — figures that pair with the up-to-30% scan-time reduction in the Philips press release. A 30% scan-time cut on a $89,340 specialist is real recovered capacity, applying that release figure to the BLS wage.

The market context underneath it is a labor squeeze, not a hardware shortage: vendors are clearing measurement-automation features now because buyers, stretched on staffing, are finally ready to pay to recover specialist hours. The Elevate Plus clearance is a direct response to that demand.

What is automated vs. what is not

Exam elementBefore Elevate PlusAfter Elevate Plus
Routine abdominal measurementManual caliper placementAuto Measure, 93%+ accuracy
Breast lesion scoringManual Bi-RADS readKoios Bi-RADS, under 2 seconds
Thyroid nodule scoringManual Ti-RADS readKoios Ti-RADS, 350,000-case library
Final diagnosis & sign-offClinicianClinician (unchanged)

Capability snapshot, by the numbers

The point of a snapshot is to keep the verified figures in one place so a buyer can sanity-check the marketing against the source. Every Philips figure traces to the Philips press release; the workforce figures to the BLS; the Koios development figure to AuntMinnie.

MetricFigureSource
Auto Measure accuracy93%+ vs. manualPhilips
Scan-time reductionUp to 30%Philips
Koios Bi-RADS read timeUnder 2 secondsPhilips
Koios Ti-RADS library350,000+ casesPhilips
Sonographer demand growth13% (2024-2034)BLS
Annual sonographer openings~5,800/yrBLS
Koios development sites48 worldwideAuntMinnie

Who shipped it

Royal Philips (NYSE: PHG) shipped Elevate Plus as a software upgrade for hardware customers already own — the EPIQ Elite and Affiniti families — rather than a new machine, as the Philips press release describes. The Koios decision-support component is licensed AI from Koios Medical, an engine the AuntMinnie report says was developed using data from 48 global sites.

The "software upgrade, not new cart" detail is the commercially important one. It means the path to ultrasound-AI for most departments is a regulated update to an installed base, which is far cheaper and faster than a capital purchase — and it is why this clearance matters to small practices, not just academic centers.

The honest limits

This is not whole-exam autonomy. Auto Measure covers specific routine measurements (abdominal, at launch), not every measurement a sonographer takes. The Koios layer scores breast and thyroid lesions against published standards; it does not replace the radiologist's diagnosis or the final sign-off, which the Philips press release leaves with the clinician.

The 93% accuracy figure is also a comparison to manual expert measurement, not a claim of perfection — which means roughly 7% of automated measurements still differ from the expert baseline, per the Philips press release. That is exactly why the human stays in the loop. And the on-cart speed gains depend on the AI output flowing cleanly into the reporting and PACS systems behind the machine — the part vendors rarely demo.

Signal vs Speculation

Everything above this line is sourced fact: the clearance date, the 93% accuracy, the up-to-30% scan-time reduction, the Koios under-2-second Bi-RADS read, and the 350,000-case Ti-RADS library, all from the Philips press release, plus the BLS workforce projections. Below is forecast.

Our read: if the 30% scan-time figure holds outside the launch sites, ultrasound-AI becomes a throughput tool first and a diagnostic tool second. The buyer's real question over the next 12-36 months is not "is the AI accurate?" but "how many more exams can I run with the staff I have?" Given the BLS-documented 13% projected growth in sonographer demand through 2034 against a thin labor pool, we expect on-cart measurement automation to move from premium add-on to default expectation on mid-range carts — the same path electronic measurement itself took.

Our read: the consolidation point will be reporting, not imaging. Once an exam produces structured Bi-RADS/Ti-RADS output during the scan, the value migrates to whatever routes that output into the EHR, the referral, the patient letter, and the billing code. That is the layer where small and mid-size practices will spend the next two years, and where teams running US Tech Automations workflows can attach the structured exam output to downstream routing without rebuilding their stack. We would bet against any forecast that treats the cart as the finish line; the cart is now an input.

How small and mid-size practices should think about it

The mistake is to read "AI ultrasound clearance" as a hardware story. For a practice that already owns an EPIQ Elite or Affiniti, this is a software decision with a labor-recovery business case. The right framing is: which routine exams eat the most sonographer time, and how much of that can a 30% cut return to the schedule?

The second framing is downstream. A faster, more standardized exam only pays off if the report moves. Practices that have mapped how an exam result becomes a referral, an authorization, and a bill will capture the gain; practices that still re-key results by hand will watch the saved scan minutes evaporate in the back office.

The third framing is consistency, and it is the one buyers tend to undervalue. The headline numbers are about speed, but the quieter benefit is that two operators now produce the same measurement on the same patient. According to the Philips press release, the explicit goal of the release is to "standardize routine exams across different operators." For a multi-site group, that standardization is worth as much as the 30% time saving, because it removes a source of variability that has historically been impossible to manage with training alone. A practice that has a senior sonographer on one shift and a new hire on another no longer ships two different quality levels — the floor rises to the AI baseline regardless of who is scanning. Teams that have already standardized their reporting through US Tech Automations workflows will find the cart now feeds that same consistent pipeline, which is the point: the standardization on the cart and the standardization in the back office reinforce each other.

None of this removes the change-management work. Sonographers have to trust the automated measurement enough to use it, which means a validation period where they check the AI against their own calipers until the 93% accuracy figure feels real in their own hands rather than in a press release. The practices that budget for that trust-building period — rather than assuming day-one adoption — are the ones that actually realize the time savings.

Key Takeaways

  • Ultrasound-AI is on-cart automation of measurement and lesion scoring; Philips Elevate Plus received FDA 510(k) clearance on June 2, 2026.

  • The hard numbers: over 93% measurement accuracy and up to 30% faster scans, per the Philips press release.

  • The "why now" is a documented sonographer-supply squeeze — 13% projected demand growth through 2034 per the BLS.

  • It ships as a software upgrade to an installed base, not a new machine — which is why it reaches small practices, not just academic centers.

  • The value will migrate downstream to whatever routes the structured exam output; that is where to invest next.

Frequently Asked Questions

What is ultrasound-AI?

Ultrasound-AI is software that automates routine ultrasound steps — primarily measurement and lesion classification — directly on the cart during the exam. In the Philips Elevate Plus clearance, it covers automated abdominal measurement plus Koios breast and thyroid scoring, as the Philips press release details.

How accurate is the automated measurement?

It reaches over 93% accuracy compared with manual measurements by clinical experts, per the Philips press release. That means a human still reviews, because roughly 7% of automated measurements differ from the expert baseline.

Does ultrasound-AI replace the sonographer or radiologist?

No. It automates routine measurement and produces a structured lesion score, but final diagnosis and sign-off stay with the clinician, as the Philips press release states. It is a capacity tool, not an autonomy tool.

Why is this happening now?

Because the workforce is the constraint. According to the BLS, sonographer demand is projected to grow 13% from 2024 to 2034 with about 5,800 annual openings, and the Philips clearance targets that gap with up to 30% faster scans.

Do I need to buy a new ultrasound machine?

Not necessarily. Elevate Plus is a software upgrade for existing EPIQ Elite and Affiniti systems, as the Philips press release makes clear, which is why it can reach practices without a capital purchase.

How big is the AI medical imaging shift?

The shift is driven by labor economics: the BLS median sonographer wage of $89,340 (May 2024) against 13% projected demand growth means every recovered hour matters, which is why the up-to-30% scan-time reduction in the Philips clearance is the feature buyers want now.


The ultrasound cart just became an input to a larger workflow. If you want to see how the structured exam output plugs into referral, authorization, and reporting automation, explore our agentic workflow platform — or start with the operator-level breakdown in what ultrasound-AI means for healthcare practices.

Freshness note: this analysis is current as of June 2026, anchored to the June 2, 2026 Philips Elevate Plus FDA clearance.

Tags

ultrasound AIAI medical imagingdiagnostic automationKoiossonography workflow

About the Author

US Tech Automations Team
AI Automation Specialists

We design agentic automation workflows for healthcare operations, imaging departments, and back-office administration.

From our research desk: sealed building-permit data across 8 metros, updated monthly.