Rad Tech Six Figures: The Salary Truth Nobody Tells You

It's Tuesday, June 9th, 2026 — you're listening to Beyond the Scan. I'm Jackie, and today we've got four stories that all, in different ways, come back to the same question: who's responsible, and what's it worth?

Quick thanks to RadiologyJobs — if you're hiring or looking, they get the imaging field in a way that general job boards just don't. Check them out.

Today: a cleared scanner, a compensation deep-dive, and two pieces about AI liability that I think belong in the same conversation.

Let's start with hardware. The FDA has cleared Siemens Healthineers' photon-counting CT scanner, the Naeotom Alpha. If you've been tracking this technology, you know this has been a long time coming. Photon-counting CT detectors work differently from conventional energy-integrating detectors — instead of averaging the energy of incoming photons, they count them individually and measure their energy level. The result is sharper spatial resolution, better spectral data, and in many configurations, lower radiation dose.

That combination — more information, less dose — is the kind of thing that takes years to actually land in a cleared device.

My read here is that this clearance matters more for what it signals than for immediate widespread adoption. Photon-counting CT hardware carries significant capital cost, and most departments aren't replacing scanners on a whim. But for the academic centers and high-volume specialty sites that have been watching this space, this clears a major regulatory hurdle. The residents doing cardiac CT, the techs running vascular cases — this is the direction the technology is moving. Worth knowing. Worth watching to see how the adoption curve actually looks over the next two years.

On the workforce side — and this is a story I think a lot of working techs will want to hear — there's a detailed breakdown of radiologic technologist compensation in 2026. The data comes from BLS figures, and a few numbers jumped out at me.

The median MRI tech salary sits at $88,180, compared to $77,660 for general radiography. That's a $10,520 annual gap. When you're thinking about whether to put in the time for an additional credential, that number makes the calculus pretty concrete.

CT adds roughly five to seven thousand dollars annually. Cardiac-interventional adds eight to twelve thousand. Travel MRI techs — and this one surprised me — can land between $100,000 and $180,000 depending on assignment and location.

Geography still matters enormously. California tops the state list at a mean annual wage of $95,070. But the article makes a point worth sitting with: states like Texas and Tennessee offer salaries in the $65,000 to $80,000 range with housing costs 30 to 50 percent below California's. The real purchasing power often comes out comparable.

For the tech on your team wondering whether to pursue MRI certification, or the one considering their first travel contract — the data here is solid and specific. Roughly 10 percent of radiologic technologists nationally earn $99,000 or more. The paths are add a modality, travel, work differentials, or move into supervision. That's the map.

Now, the AI liability conversation — and there are two pieces this week that I think you need to read together.

The first, from Boles Blogs, digs into a proposal for AI-only radiology reading where radiologists would only confirm flagged abnormalities. The author's core argument leans on data from Kristina Lång and colleagues at Lund University — research published in The Lancet Oncology in 2023 and followed up in The Lancet in 2026. The AI-assisted arm of that trial reduced radiologist workload by 44 percent and showed a 12 percent reduction in interval cancers compared to standard double reading.

Here's what the article is careful to point out: that evidence base supports AI-assisted reading, where a human radiologist stays in the loop throughout. A system where radiologists only see the cases the algorithm flags as abnormal creates a liability gap with no clear defendant when the algorithm is wrong and no human ever looked.

My take: the Lund data is genuinely strong. But it describes a human-in-the-loop model, and proposals that use it to justify full AI autonomy are reading past what the evidence actually shows.

Which connects directly to the second piece, from Dr. Gigi Magan's Substack. This one is called "Liable Either Way," and the title says it. A 2025 systematic review of 83 studies found no overall performance gap between AI and physicians — but with substantial variability by condition and specialty. Meanwhile, malpractice risk is now forming from both directions: using AI and not using it. No formal guidance exists. No training covers it. The clinician holds the liability in both scenarios.

That's the situation right now. The research is accumulating in multiple directions simultaneously, and the people doing the actual reading are exposed on every side. I want to be careful here, because this isn't cause for panic — but it is cause for your department to have a documented, explicit policy on AI use before a case ever goes sideways. If you don't have that conversation yet, this is the week to start it.

Alright — a cleared photon-counting scanner, a compensation map for techs who want to plan ahead, and a liability picture that is genuinely complicated and getting more so. That's the episode.

Thank you for spending part of your day here. If the show is useful to you, share it with a colleague — that's still the best way to grow this. You can find us at beyondthescan.io. I'll be back in two weeks with more updates.