Industry Brief  ·  Radiology

Your reads are research.

The federal R&D tax credit pays radiology groups for testing new reading tools and protocols. Most groups never claim it.

Book a call

It costs nothing to find out. If there isn't enough credit to be worth claiming, you don't pay a thing.

New to youis enough. A reading tool or protocol your group hasn't used before can count when you had to test your way to it.
4 years of creditin one study: this year, plus past years still open to amend, usually three.
Exam supportcomes with every study. If the IRS asks, we answer.

What counts in your group

Work your readers already do
  • AI second-read validationReading alongside a tool and logging where its score disagrees with yours.
  • Protocols for new capture methodsDeciding how to read a reconstructed automated ultrasound set, and when to recall.
  • Comparison priorsGetting years of outside priors in front of a reader or tool to spot change.
  • Handheld versus automatedFinding which regions read better by hand, then changing the protocol.
  • Reader-level outcomesUsing reader-level outcomes to test a new protocol.
  • Vendor beta programsBuilding your own protocol around a vendor's beta release.
  • Algorithm revalidationRe-reading your data set each time a vendor reconfigures the algorithm.
  • Cloud reading platformsDesigning how your reads run on a new cloud platform when nobody knew if it would hold.

The four questions, for radiology groups

§41's four-part test
  1. Did you bring in a reading tool or protocol new to your group?It doesn't have to be new to the specialty.
  2. Did it take real know-how?Your reads rest on biology and imaging physics; the tools on computer science.
  3. Were you unsure at the start?Does the algorithm flag the right region at the right depth? Your group can't know until your readers test it.
  4. Did you try options to find out?Scoring every read against the tool and acting on the mismatches is exactly that.

Unsure about any of these? That's what the call is for. The four-part test in plain English

Radiology groups ask

Isn't the AI tool the vendor's research?

The product is theirs. Whether it works on your studies, against your readers, at your sites is a question only your group can answer.

The normal read still happens. What counts on top of it?

The added work: scoring the tool, chasing down disagreements, and revising the protocol. The standard read itself doesn't count.

Do our peer review and QA records help?

It's some of the best evidence there is. Outcomes by reader and by site show the experiment and its result.

We read for sites in several states. Does that matter?

State credits differ, and the study checks each one. Where the work happened decides which state's rules apply.

Filing, how far back, exams, and costs: common questions.

For your CPA: where the credit goes on Form 1120, 1120-S, 1065, or 1040.

Other industry briefs

Where else it lands
Diagnostic Imaging

Every new modality is a trial.

Testing how new modalities, AI reads, and screening protocols work for your patients can qualify.

Read the brief Medical Imaging

Your image pipeline is research.

Building interfaces and integrations nobody sold you can be qualified development.

Read the brief Medical Practices

Your protocols are research.

Building and testing the protocols around new technology in patient care can qualify.

Read the brief

Give us thirty minutes.

Bring the radiologist who runs your tool evaluations. You'll leave knowing whether your group's work qualifies.

Book a call Not ready to talk? See if you’re a fit in two minutes.