What "self-interpretation" actually means
Self-interpretation is what happens when the physician who orders an imaging study — an X-ray, ultrasound, CT, MRI, or nuclear medicine scan — also personally reads and reports on it, instead of sending it to a radiologist. It's legal: a carve-out in the federal Stark law, the in-office ancillary services exception, allows physicians to bill for imaging performed and interpreted within their own practice, which was originally meant to permit convenient, same-visit ancillary care but in practice has become a wide opening for self-referral.
It's also more common than most people outside radiology assume. A Harvey L. Neiman Health Policy Institute analysis of 1.6 million office-based imaging claims from a 2022 Medicare 5% research file, published in the American Journal of Roentgenology, found ordering providers self-interpreted 43.6% of the imaging they ordered, and interpreted 58.5% within their own practice (by themselves or a colleague) rather than sending it to an outside radiologist.
It skews heavily toward the modalities easiest to misread
The rate isn't uniform. Self-interpretation clusters in the modalities that require the least equipment investment and, not coincidentally, are the easiest for a non-radiologist to misjudge without subspecialty training — while cross-sectional imaging is almost always sent to a radiologist:
| Modality | Self-interpretation rate |
|---|---|
| Ultrasound | 52.0% |
| Radiography / fluoroscopy | 50.4% |
| Nuclear medicine | 39.5% |
| MRI | ~6.1% |
| CT | ~5.2% |
By specialty, the pattern is just as stark: 75.7% of orthopedic and sports-medicine orders and 73.0% of cardiology orders were self-interpreted, compared with 30.5% for nonphysician practitioners and 19.9% for primary care physicians. Having a radiologist on staff mattered — the adjusted odds of self-interpretation were roughly half for providers with a within-practice radiologist compared with those without one, across most specialty-modality combinations studied.
The researchers were careful about what this does and doesn't show: nonradiologist self-interpretation "warrants scrutiny" precisely because training in image interpretation for most other specialties is narrower in scope and shorter in duration than a radiology residency and fellowship — not because every self-read is wrong.
The newer finding: self-interpreters also order more imaging
What turned this from a familiar quality concern into fresh news is a follow-on analysis, published in the Journal of the American College of Radiology and reported by Radiology Business, that connects self-interpretation to how much imaging gets ordered in the first place. Providers who self-interpreted 100% of their own imaging orders had roughly 2.88 times higher adjusted odds of ordering imaging for their patients than providers who didn't self-interpret at all — a spread that ranged from about 1.54 times higher for MRI to 3.51 times higher for ultrasound.
Study co-author Dr. Vijay M. Rao, former radiology chair at Thomas Jefferson University, framed the point carefully rather than as a turf argument: "who interprets imaging may influence clinical decision-making in important ways," and the concern is not about barring non-radiologist specialists from ever interpreting a scan. ACR CEO Dr. Dana Smetherman put the utilization link more directly, noting the study identifies self-interpretation as a factor associated with higher imaging use overall — the kind of finding that turns an interpretation-quality debate into a cost and appropriateness debate too.
What "stricter rules" would actually target
Two policy threads follow from this research, and they attack different parts of the problem:
Revisit the in-office ancillary exception
The Stark law was written to prevent physicians from profiting off referrals to services they have a financial stake in — but its in-office ancillary carve-out is exactly what makes imaging self-referral legal. The Neiman researchers point to narrowing that exception, or at minimum improving patient disclosure of who is actually interpreting a scan, as the most direct fix.
Restore decision support at the point of ordering
Separately, ACR is backing the Radiology Outpatient Ordering Transmission (ROOT) Act, folded into the bipartisan Patients First Act, which would revive Medicare's appropriate-use-criteria consultation requirement for advanced imaging. ACR projects it could save Medicare roughly $2 billion over ten years — a lever aimed at order volume rather than who reads the result.
Neither proposal is close to law yet — the ROOT Act's fate depends on the broader Patients First Act, and no Stark reform bill specific to imaging has been introduced. But the fact that a peer-reviewed policy analysis is now tying self-interpretation directly to utilization, not just accuracy, changes the argument radiology groups can make to lawmakers and payers. (For the mechanics of the ROOT Act and what it would change for imaging orders more broadly, see our Patients First Act explainer.)
Why self-interpretation persists — and what removes the incentive
Self-interpretation rarely happens because a practice wants to avoid a radiologist's opinion. More often it persists because getting a fast, reliable radiologist read is friction-heavy: a referring orthopedic or cardiology practice weighs the delay and cost of sending a study out against reading it themselves and moving the patient along the same day. That's the gap AI-assisted CT reporting is built to close. A structured report draft generated quickly, reviewed by xAID's in-house radiologist, and delivered ready-to-sign to the practice's own radiologist gives referring physicians and imaging centers a fast, compliant alternative — one that keeps a trained radiologist in every read without the turnaround delay that made self-interpretation tempting in the first place.
Frequently asked questions
What is self-interpretation of imaging orders?
Self-interpretation happens when the physician or provider who orders an imaging study — an X-ray, ultrasound, CT, MRI, or nuclear medicine scan — also personally interprets it, rather than referring it to a radiologist. It is legal under a federal exception, but it removes the independent check a radiologist normally provides.
How often do non-radiologist providers self-interpret the imaging they order?
A Harvey L. Neiman Health Policy Institute analysis of 2022 Medicare claims, published in the American Journal of Roentgenology, found ordering providers self-interpreted 43.6% of office-based imaging studies overall. Rates were highest for radiography and fluoroscopy (50.4%) and ultrasound (52.0%), and lowest for CT (about 5%) and MRI (about 6%). By specialty, orthopedic and sports-medicine providers self-interpreted 75.7% of their own orders and cardiologists 73.0%, versus 19.9% for primary care physicians.
Does self-interpretation lead to more imaging being ordered?
A newer Neiman Health Policy Institute analysis published in the Journal of the American College of Radiology found providers who self-interpreted 100% of their own imaging orders had about 2.88 times higher adjusted odds of ordering imaging for their patients than providers who didn't self-interpret, ranging from roughly 1.54 times higher for MRI to 3.51 times higher for ultrasound.
What policy changes are being proposed to address self-interpretation?
Researchers have suggested revisiting the Stark law's in-office ancillary services exception, which currently allows self-referral for imaging that would otherwise be restricted, and improving patient awareness of who is interpreting their scan. The American College of Radiology is also backing the ROOT Act, folded into the bipartisan Patients First Act, which would restore point-of-order clinical decision support for advanced imaging — a measure ACR projects could save Medicare about $2 billion over ten years.
Source: Harvey L. Neiman Health Policy Institute analysis, "Self-Interpretation of Imaging Studies by Ordering Providers," American Journal of Roentgenology (2025), doi.org/10.2214/AJR.25.32769; follow-on analysis published in the Journal of the American College of Radiology, as reported by Radiology Business. Figures are rounded as reported.