Skip to content
Manufacturer-independent. Connects over HDMI.

Explainable TI-RADS for any ultrasound.

Diagnoflux reads your ultrasound's HDMI output and overlays explainable TI-RADS risk in real time. No new hardware.

A Diagnoflux cart: a second monitor shows a dual-plane thyroid ultrasound with lime segmentation outlines and risk badges, above the Diagnoflux gateway box
Live overlay 10-20 Hz

Backed and validated by

Supported by TÜBİTAK Built at ODTÜ Teknokent Clinical validation with Akdeniz University Partner: Kardinero Medical Systems

Ultrasound is operator-dependent, so decisions drift.

Thyroid risk depends on who holds the probe and which device they use. The same nodule can be scored differently across readers and machines, while most real-time AI stays locked inside a few premium systems.

How it works

From an HDMI cable to explainable risk, in real time.

A small gateway sits between your ultrasound and a second monitor. It adds an AI layer on top of the device you already own, without touching its certified hardware or software.

The Diagnoflux gateway: front panel with HDMI In, HDMI Out, Ethernet, two USB ports and a power button
The Diagnoflux gateway. HDMI in, HDMI out, Ethernet. About the size of a small router.
  1. 01

    Connect

    Plug the gateway into your ultrasound's HDMI output and a second monitor into its HDMI out. Nothing changes on the ultrasound itself.

  2. 02

    Analyze

    Frames stream over an encrypted link to a single multi-head AI model, running in a Turkey-based cloud or on your own on-prem server.

  3. 03

    Overlay

    Segmentation and TI-RADS overlays appear on the second monitor in real time. A structured report is generated after the exam.

What it does

Explainable TI-RADS, broken down.

Diagnoflux follows the same TI-RADS workflow a clinician already uses, then shows its working at every step instead of returning a single opaque score.

The Diagnoflux interface: a dual-plane thyroid ultrasound with nodule and gland segmentation, a per-component TI-RADS breakdown, and a TR4 result at 82 percent confidence
A live thyroid exam: segmentation, per-component scoring and the TR level with calibrated confidence. Segmentation 10-20 Hz Components 2-3 Hz
Component level

Five TI-RADS components, scored

  • Composition
  • Echogenicity
  • Shape
  • Margin
  • Echogenic foci

Two planes, one TR level

You mark the representative frame in the transverse and longitudinal planes. Each component is scored and combined into the ACR TI-RADS level.

TR1 TR2 TR3 TR4 TR5

Calibrated, not a black box

Every result carries a per-component breakdown and a calibrated confidence, so you can see what drove the score rather than trusting a single number.

Reports that fit your workflow

After the exam, findings export to PDF and DICOM, ready for PACS, with the overlays and component scores attached.

PDF DICOM PACS-ready

Device-independent changes the math.

Because Diagnoflux sits on top of existing equipment rather than replacing it, the value shows up in cost, reach and day-to-day clinical work.

Works with the devices you own

Any ultrasound with an HDMI output joins in. No rip-and-replace, and no vendor lock-in to a single manufacturer.

Lower total cost of ownership

Upgrade through a gateway and a license instead of new capital equipment, with predictable running costs.

Fewer avoidable biopsies

Standardized, explainable reads aim to cut unnecessary fine-needle biopsies and even out follow-up between centers.

No new workflow

There is no second complex interface. Clinicians scan as usual while the overlays appear on the monitor alongside.

Security and deployment

Your data stays where your policy needs it.

Run Diagnoflux in a Turkey-based cloud or entirely inside your own network. Either way, the link is encrypted and access is controlled and logged.

Turkey-based cloud

Managed GPUs hosted in Turkey. Frames are encrypted in transit and processed under data rules built for KVKK and GDPR.

  • TLS 1.3 encrypted transport
  • Short-lived JWT authentication
  • Role-based access and audit logging

End-to-end latency target P95 ≤ 150 ms

On-premise

The analysis server runs inside your own network, so imaging data never leaves the building. Ideal where data residency is non-negotiable.

  • Runs on your servers and network
  • No data leaves your premises
  • Your retention and backup policy

End-to-end latency target P95 ≤ 80 ms

Built to the right standards

Software lifecycle and risk management follow IEC 62304, ISO 13485 and ISO 14971. The software is assessed under MDR 2017/745 Rule 11, with the gateway documented as a non-diagnostic accessory.

  • MDR 2017/745
  • IEC 62304
  • ISO 13485
  • ISO 14971
  • KVKK
  • GDPR

Diagnoflux is an investigational device undergoing clinical validation and is not yet CE-marked. Regulatory references describe the development pathway, not a completed certification.

Built for the people in the room.

The pilot focuses on thyroid endocrinology, but the value reaches everyone who relies on the same scan, from the clinic to the quality office.

Endocrinology and radiology

Physicians performing thyroid ultrasound day to day.

Hospitals and imaging centers

Public, private and university sites, and hospital chains.

Quality and IT teams

Standardized reporting, data residency and access control.

Training programs

A shared visual reference for less-experienced readers.

One platform, beyond thyroid

The same gateway works with any HDMI optical or video system, so the approach extends well past the thyroid pilot.

  • Endoscopy
  • Colonoscopy
  • Cystoscopy
  • Bronchoscopy
  • Dermatoscopy
  • Breast US
  • Carotid US
  • Musculoskeletal US
  • Liver US

Questions, answered.

The things clinical, IT and procurement teams ask first. Need something more specific? Ask the team.

See Diagnoflux on your own devices.

Tell us about your setup and we will arrange a walkthrough or an on-site pilot. We typically reply within two business days.

Request a demo

Fields marked with * are required.

We use your details only to respond to this request. Notrino Research, ODTÜ Teknokent, Ankara.