Softwr

Healthcare · head to head

Rad AI vs Therapixel

Rad AI logo

Rad AI

Healthcare

Generative reporting and follow-up tracking for radiology, positioned outside the diagnostic device boundary

From
On request
Rated
-
Full reviewWebsite
Therapixel logo

Therapixel

Healthcare

French mammography AI with FDA clearances for 2D and tomosynthesis breast screening

From
On request
Rated
-
Full reviewWebsite

The short version

  • Each has a real cost: Rad AI the products are not FDA cleared devices and are not intended to be, so any claim that Rad AI improves diagnostic accuracy is unsupported; what has been measured is reporting speed and follow-up completion, not detection.; Therapixel european and American regulatory positions were established under different regimes, and under the EU Medical Device Regulation this class of diagnostic AI faces a higher classification and notified body scrutiny, so European buyers must verify the current MDR certificate and expiry rather than relying on a legacy CE reference.
  • They diverge on capability: Rad AI covers Impressions, Therapixel covers MammoScreen 2D.
  • Prices and features above were last checked on 1 September 2026.

Where they differ

Only the attributes on which Rad AI and Therapixel actually diverge.

Attributes where Rad AI and Therapixel differ
AttributeRad AITherapixel
PlatformsWindows, Web, CloudWeb

Identical on both: starting price (On request), pricing model (quote), free tier (No), user rating (Not yet rated), category (Healthcare).

What each one covers

Drawn from each product's published feature list. An absence here means we hold no record of it - not that the product lacks it.

Only in Rad AI

  • Impressions
  • Reporting
  • Continuity
  • Style learning
  • Omni Reconcile
  • PACS and dictation embedding
  • Patient outreach workflows
  • Audit reporting

Only in Therapixel

  • MammoScreen 2D
  • MammoScreen DBT
  • One to ten suspicion score
  • Summary report
  • Workstation and PACS delivery
  • Reading workflow configurations

What people use each for

The jobs each tool is most often brought in to do.

Rad AI

  • A private radiology practice whose read volume has grown faster than it can recruit and that wants dictation time per study reduced without changing PACSnot Therapixel
  • A health system losing revenue and exposing itself to liability because recommended follow-up imaging on incidental findings is never bookednot Therapixel
  • A radiology department standardising impression quality across a mixed group of employed and locum radiologistsnot Therapixel
  • A group deploying an AI tool that must not trigger a device validation programme, because governance capacity is already consumed by a triage algorithm rolloutnot Therapixel

Therapixel

  • A European breast screening programme running mandatory double reading that wants to evaluate AI as a concurrent or second reader within what national guidance permitsnot Rad AI
  • An American breast centre that has moved to tomosynthesis and needs reading support on DBT specifically, which Therapixel holds a clearance fornot Rad AI
  • A screening service wanting a graded suspicion score rather than a binary flag so readers can calibrate their own thresholdnot Rad AI
  • A radiology group evaluating multiple mammography AI vendors head to head on their own historic case set before committingnot Rad AI

Where each one falls short

Documented limitations, not opinions. Every one is a constraint you would hit in normal use.

Rad AI

  • The products are not FDA cleared devices and are not intended to be, so any claim that Rad AI improves diagnostic accuracy is unsupported; what has been measured is reporting speed and follow-up completion, not detection.
  • Generated impressions require radiologist review and signature, so the time saving evaporates in any group whose radiologists rewrite the draft rather than edit it, and adoption varies sharply by individual.
  • It is bound to your dictation platform, so a group planning to move off PowerScribe should sequence that decision first or expect to redo the Rad AI integration.
  • Continuity depends on the quality of recommendation language in historical reports, and departments with inconsistent phrasing get poor recall until reporting practice is standardised.
  • Pricing is not published and is structured per radiologist or per study depending on the product mix, which makes it hard to compare against image-analysis AI vendors that price per scan.

Therapixel

  • European and American regulatory positions were established under different regimes, and under the EU Medical Device Regulation this class of diagnostic AI faces a higher classification and notified body scrutiny, so European buyers must verify the current MDR certificate and expiry rather than relying on a legacy CE reference.
  • MammoScreen does mammography only, so a hospital wanting AI across head CT, chest and breast needs several vendors and several PACS integrations, or an orchestration layer bought from someone else.
  • Therapixel is a small French company competing against Hologic and iCAD, which have incumbency inside the mammography reading workstation, so integration into an American breast centre is more work than it is for the workstation vendor own AI.
  • Detection performance is sensitive to the mammography unit vendor and acquisition protocol, so published results do not transfer between estates and each site needs local validation it must fund itself.
  • Whether AI can substitute for one of two human readers is a national workforce and liability question, not a product capability, so the labour saving that justifies the purchase in one country is not permitted in another.

Pricing, plan by plan

Rad AI

On request
  • Rad AI platform$undefined/year
    • Priced per radiologist or per study volume depending on product and contract
    • Impressions, Reporting and Continuity licensed separately
    • Annual subscription with implementation and integration scoped per site

Therapixel

On request
  • MammoScreen$undefined/year
    • Priced per examination or by annual screening volume
    • 2D and tomosynthesis licensed by the versions cleared for each
    • Cloud or on premises deployment quoted differently

Which should you pick?

Choose Rad AI if

  • You need impressions.
  • You work on Windows, Web, Cloud.
  • You also want reporting.
Rad AI review

Choose Therapixel if

  • You need mammoscreen 2d.
  • You also want mammoscreen dbt.
Therapixel review

Questions people ask

Is Rad AI or Therapixel better?
Neither clearly leads. Rad AI starts at On request and Therapixel at On request, and user ratings are close enough to be indistinguishable. Choose on capability and platform support.
Which is cheaper, Rad AI or Therapixel?
Rad AI starts at On request and Therapixel at On request.
Does Rad AI or Therapixel run on more platforms?
Rad AI runs on Windows, Web, Cloud. Therapixel runs on Web.
What is Rad AI best used for?
Rad AI is most often used for a private radiology practice whose read volume has grown faster than it can recruit and that wants dictation time per study reduced without changing pacs, a health system losing revenue and exposing itself to liability because recommended follow-up imaging on incidental findings is never booked, a radiology department standardising impression quality across a mixed group of employed and locum radiologists, a group deploying an ai tool that must not trigger a device validation programme, because governance capacity is already consumed by a triage algorithm rollout. Of those, a private radiology practice whose read volume has grown faster than it can recruit and that wants dictation time per study reduced without changing pacs and a health system losing revenue and exposing itself to liability because recommended follow-up imaging on incidental findings is never booked are not what Therapixel is typically brought in for.
What can Rad AI do that Therapixel cannot?
Rad AI covers Impressions, Reporting, Continuity, Style learning. Therapixel covers MammoScreen 2D, MammoScreen DBT, One to ten suspicion score, Summary report.

Answered from the vendors’ own pages

Rad AI: Is Rad AI FDA cleared?

No, and by design. The products draft text and manage follow-up workflow rather than interpreting images, so they fall outside the device pathway. Treat any diagnostic accuracy claim with scepticism.

Therapixel: Is MammoScreen FDA cleared?

Yes. It holds 510(k) clearance K192854 for two dimensional mammography from July 2020, a November 2021 clearance for digital breast tomosynthesis, and two further clearances in October 2024 for later versions.

Rad AI: Does the radiologist still sign the report?

Yes. Output is a draft that the radiologist reviews and signs; nothing is released autonomously.

Therapixel: Is it CE marked?

Yes, but European buyers should ask specifically for the current certificate under the Medical Device Regulation and its expiry date, because this class of diagnostic AI moved to a higher classification under MDR.

Rad AI: Which product actually pays for itself?

Continuity has the clearest financial case, because recommended follow-up imaging that gets completed is billable work that currently does not happen.

Therapixel: What does the one to ten score mean?

It is a graded suspicion score per finding and per study, designed to be integrated alongside a radiologist BI-RADS assessment rather than treated as a positive or negative result.

Rad AI: Will it work with our dictation system?

It is deepest with Nuance PowerScribe. Confirm your specific dictation and PACS combination before contracting, because integration depth is the main deployment variable.

Therapixel: Does it cover anything besides mammography?

No. It is a mammography product only, including tomosynthesis. Other body regions need a different vendor.

Share

Related pages

More on Rad AI

More on Therapixel

Keep looking

Other head to heads