Samsung Medison has launched the HERA Z10, a premium ultrasound system for obstetrics and gynaecology that brings AI-assisted measurements, fetal imaging and workflow automation into a smaller sibling of its flagship HERA Z20. The company presented the system at the International Society of Ultrasound in Obstetrics and Gynecology congress in London, held September 4–7.
The HERA Z10 matters because it applies software automation to repetitive parts of an ultrasound examination while keeping interpretation with trained clinicians. Samsung says availability and features can vary by country, and its performance figures should be read as company-reported workflow results rather than proof of improved diagnosis or patient outcomes.
| Item | Reported detail |
|---|---|
| Clinical focus | Obstetrics and gynaecology ultrasound |
| Live ViewAssist claim | Up to 52% less time for a mid-trimester detailed scan in a joint study |
| EzStructure claim | Up to 71% less manual manipulation in a joint study |
| Other tools | MV-Flow 3D and FibroidVue |
| Launch venue | ISUOG congress in London, September 4–7 |
What the HERA Z10 is designed to do
The system targets women’s-health imaging, including fetal assessment and gynaecological examinations. It inherits AI workflows and beamforming technology from the HERA Z20. Beamforming controls how ultrasound signals are transmitted and combined to create an image; improvements can affect clarity and consistency, but image quality remains dependent on the patient, probe, operator and examination.
Samsung also introduced neonatal probes for imaging small patients and difficult anatomical regions. The launch broadens the HERA family rather than replacing the Z20: the Z10 offers a premium system at a different footprint and configuration.
Live ViewAssist automates selected measurements
Live ViewAssist is designed to recognise required fetal planes during a live scan, capture standard views and generate selected measurements. Samsung says a joint study found it could reduce the time required for a mid-trimester detailed scan by as much as 52%. The wording “up to” is important: it describes the best reported reduction under the study conditions, not a guaranteed saving for every patient, operator or clinic.
Automation can reduce the manual sequence of freezing a frame, selecting a measurement tool and placing callipers. It may also make a repeatable protocol easier to follow. But a saved interaction is not the same as a clinical conclusion. A sonographer or physician still has to confirm that the correct plane was captured, that landmarks are visible and that a measurement is plausible.
EzStructure and anatomy identification
EzStructure is intended to identify fetal anatomical structures automatically. Samsung reports that it reduced manual manipulation by up to 71% in a joint study. As with the scan-time figure, the release does not establish a universal result. The relevant questions for hospitals include the study population, operator experience, protocol, comparator and frequency of corrections.
The practical value of structure recognition is consistency. Detailed fetal examinations require a set of standard images, and missed or duplicated views can extend an appointment. A system that organises recognised views could make review easier, particularly in high-volume departments. It should not be understood as autonomous anomaly detection unless a specifically cleared function and supporting evidence say so.
Tools for vascularity and fibroids
MV-Flow 3D visualises microvascular blood flow in three dimensions. Vascular information can support assessment of tissue and lesions, but its meaning depends on the indication and clinician’s interpretation. The feature is an imaging tool, not a diagnosis on its own.
FibroidVue segments uterine fibroids and produces measurements. Fibroids can be numerous and irregularly shaped, so software-assisted segmentation may reduce repetitive outlining and help document changes. A clinician still needs to verify boundaries and connect the measurements to symptoms, fertility goals and treatment planning.
Why workflow AI is the central product story
Medical-imaging AI is often described as detection or diagnosis, but the HERA Z10 announcement is mainly about workflow: recognising planes, automating measurements, organising structures and segmenting anatomy. These are narrower tasks with clearer points for human review. They can be commercially valuable because ultrasound is operator-dependent and departments face time pressure.
The design also shows how AI is being packaged as part of an imaging platform. Hardware, probes, signal processing, ergonomic controls and software all influence whether a feature is usable. A technically accurate model can still fail to save time if it interrupts the examination or produces outputs that require extensive correction.
Samsung’s wider software strategy extends beyond healthcare. Our report on Samsung’s Tizen 10 appliance update looks at platform deployment across consumer hardware. For another example of non-contact measurement in advanced engineering, see KERI’s superconducting-tape inspection system.
Evidence hospitals should request
Before procurement, a hospital should ask for the full study methods behind the 52% and 71% figures, including sample size, sites, examiner experience and exclusions. It should also evaluate correction rates, performance across diverse patients, integration with reporting and picture-archiving systems, cybersecurity controls, service coverage and local regulatory status.
Local validation is especially important in ultrasound because performance can vary with body habitus, fetal position, gestational age and equipment settings. Buyers should compare a representative mix of examinations, not only ideal scans. They should also measure whether automation changes throughput without increasing rescans or downstream review.
Human oversight and accountability
The system’s automated outputs need a visible review path. Clinicians should be able to inspect the source image, adjust a measurement and understand which functions were used. Clinics also need procedures for software updates, because a new version can change behaviour even when the hardware is unchanged.
Training remains essential. Automation may reduce routine actions, but users must recognise when a suggested plane or segmentation is wrong. Clear logs, quality assurance and escalation processes matter more than a single performance percentage.
What the launch does and does not establish
The announcement establishes the HERA Z10’s product positioning and feature set, supported by multiple direct reports from the launch. It also provides two quantitative workflow claims. It does not, in the cited material, demonstrate improved diagnostic accuracy, fewer adverse outcomes or performance across every clinical setting.
That boundary does not make the product insignificant. Reducing repetitive steps can matter to clinicians and patients if the gains are reproducible and safely integrated. The next useful evidence will be detailed study publications, regulatory clearances by market, independent evaluations and real-world data after deployment.
Procurement is more than a feature comparison
Ultrasound systems are long-lived clinical assets, so procurement teams should calculate the full operating burden. That includes probe replacement, preventive maintenance, software licensing, network integration, staff training and access to service engineers. An AI feature that saves minutes during a demonstration may deliver little value if it is unavailable in the purchased configuration or creates extra documentation work.
A careful evaluation should use the department’s own protocols and a representative group of operators. Experienced specialists and newer sonographers may interact with automation differently. Recording the frequency and size of manual corrections can reveal whether an apparently faster workflow simply shifts effort into later review. Departments should also track aborted automation, repeat acquisitions and cases in which the suggested output is rejected.
Patient communication still belongs to the care team
Fetal ultrasound can be emotionally sensitive. Faster acquisition should not be confused with a shorter or less careful clinical encounter, and an automated label should not be presented to a patient as a finding before clinical review. Hospitals need clear rules about when outputs appear on a display, who can explain them and how uncertain or technically limited examinations are documented.
The same applies to gynaecological tools such as FibroidVue. A segmented structure can support documentation, but treatment decisions depend on symptoms, location, growth, age and patient preference. Keeping the software’s task narrow and its output reviewable is the safest interpretation of the product announced so far.
Deployment needs measurable acceptance criteria
A responsible rollout should define success before the system enters routine use. Useful measures include acquisition time, the share of automated measurements clinicians edit, repeat-image rates, technical failures and the number of cases escalated for specialist review. Those indicators connect the vendor’s workflow claims to the reality of a particular department without pretending that speed alone represents clinical quality.
Hospitals should also establish a baseline using their existing equipment and protocol. Without that comparison, a faster examination after installation could reflect training, staffing or case mix instead of the software. A staged introduction, with reviewed cases and documented corrections, gives governance teams evidence for expanding or limiting individual features.
Frequently asked questions
Does the HERA Z10 diagnose conditions by itself?
No. The announced tools assist with image capture, measurements, structure identification, flow visualisation and segmentation. Qualified clinicians remain responsible for interpretation and diagnosis.
What does the 52% figure refer to?
Samsung says a joint study found Live ViewAssist reduced mid-trimester detailed-scan time by up to 52%. It is a maximum company-reported workflow result, not a universal guarantee.
What does the 71% figure refer to?
Samsung reports that EzStructure reduced manual manipulation by up to 71% in a joint study. Full study methods are needed to judge generalisability.
Is the HERA Z10 available everywhere?
No. Samsung states that product availability and features may vary by country. Buyers should confirm local regulatory status and configurations.
How is it different from the HERA Z20?
The Z10 extends the premium HERA range and inherits AI workflows and beamforming technology from the flagship Z20, while occupying a different product tier and configuration.
Get the day’s top stories in your inbox
One concise email. No spam, unsubscribe anytime.
