Midjourney, the AI image-generation startup, launched a medical imaging venture last week built around a water-tank ultrasound scanner that the company claims can capture a sub-millimeter 3D map of the human body in 60 seconds. CEO David Holz has suggested the system could one day surpass MRI. Four medical imaging professors interviewed by The Verge said the claims are not supported by any public evidence Midjourney has released so far.
The product, called Midjourney Medical, lowers a user into a vat of water on a platform ringed by underwater sensors that send sound waves into the body and capture the echoes. Midjourney's head of medical, Tom Calloway, said the scanner uses AI and specialized chips to handle what the company calls 'unthinkably huge amounts of data and processing power,' with AI also handling lossless compression to speed up processing. The target scan time is 60 seconds, versus the 30 minutes or more typical of conventional ultrasound exams.
Midjourney is not initially seeking FDA clearance for the device, citing the high regulatory bar and clinical trial requirements involved. Instead, the company plans to install the machines in spas, treating scans as a 'side-effect' of a visit. The marketing material cites a claim that early imaging could prevent 30% of deaths and 50% of healthcare costs, while concept renderings show gold-trimmed pools rather than clinical environments.
Key facts
- 01Midjourney announced Midjourney Medical, a submerged ultrasound scanner targeting 60-second scans versus 30+ minutes for typical ultrasounds.
- 02CEO David Holz says the system could one day rival MRI; radiologists at four US medical schools told The Verge the claims lack public evidence.
- 03The company cites a statistic that 30% of deaths and 50% of healthcare costs could be avoided with enough early imaging.
- 04Midjourney is not initially pursuing FDA clearance and plans to install the machines in spas, framing scans as a 'side-effect' of a visit.
- 05Midjourney claims sub-millimeter resolution; the University of Michigan's Venkatesh Murthy says the published images are 'decidedly low-res.'
The pivot is a sharp turn for a company whose existing business is generating synthetic images for paying subscribers. AI is barely mentioned in the company's blog post announcing the move, and the relationship between Midjourney's image-generation product and a hardware-and-imaging medical venture is not spelled out. The connective tissue, as Calloway describes it, is computational: handling the volume of raw ultrasound data the underwater sensor array would produce.
Venkatesh Murthy, a professor of preventive cardiology, internal medicine, and radiology at the University of Michigan Medical School, called the technology 'super cool' and said he was glad to see prototypes built. But he said there is 'a long road ahead' to generating high-quality images and demonstrating clinical benefit. Murthy described the published sample images as 'decidedly low-res' and said comparisons to MRI are unsupported.
“claims about resolution are clearly theoretical”— Venkatesh Murthy, Professor of preventive cardiology, internal medicine, and radiology at the University of Michigan Medical School
Mark Anastasio, a professor of imaging sciences at Washington University in St. Louis, called the project an 'exciting development' and said whole-body ultrasound prototypes have been demonstrated by researchers. He added that there 'is no current evidence' that detailed ultrasound scans of this kind can match MRI. Matthew Davenport, also at the University of Michigan, was blunter: he said Midjourney's images were 'interesting' but the claims are 'wildly unsubstantiated,' calling them 'perhaps the most grandiose' he has seen.
Calloway, responding to the MRI and CT comparisons, said the first-generation scanner is 'fast, affordable, radiation-free, and great at imaging soft tissues like muscle and fat for body composition' but acknowledged it 'won't be a replacement' for MRI or CT. That framing is narrower than the company's own marketing, which compares the resulting images directly to MRI scans and cites mortality statistics tied to diagnostic imaging.
“The first-generation scanner is fast, affordable, radiation-free, and great at imaging soft tissues like muscle and fat for body composition — but it won't be a replacement for either of these technologies”— Tom Calloway, Head of medical at Midjourney
Scott Reeder, a professor of radiology at the University of Wisconsin School of Medicine and Public Health, called the concept innovative but said it is 'not a proven technology.' He noted that ultrasound has physical limits MRI does not — air and bone block sound waves, complicating images of the gut, the pelvis, and parts of the head. Without a technician maneuvering probes, those regions become harder to capture cleanly in a single 60-second pass.
William Morrison, a professor of radiology at Thomas Jefferson University, was the most skeptical. He described the launch as a 'vibe-based rollout' and said the sample images are 'markedly limited compared to existing technology such as CT and MRI.' Morrison said the water-bath approach has been largely abandoned because of the physical constraints of ultrasound. Practical concerns mount from there: the water would need to be 'completely pure, without bubbles or dirt,' likely changed between customers, and users would probably need to be 'scrubbed and completely shaven' to avoid trapped air deflecting sound waves.
Midjourney's public evidence so far consists of a blog post, concept renderings, and a small set of phantom and body scans. There are no published comparisons against MRI from a controlled study, no peer-reviewed validation, and no clinical pilot data. The company says medical applications will come later; for now, the device is positioned as a wellness product, which sidesteps the regulatory regime that would otherwise require it to prove what it shows.
The wellness-first route is a familiar pattern in consumer health tech — ship the hardware, sell the experience, and let clinical validation catch up later, if ever. It is a viable commercial path, and Midjourney has the brand and capital to run it. But the gap between 'a 60-second scan at the spa' and 'something as powerful as MRI' is the gap radiologists are pointing at, and the company has not yet shown the data that would close it. Until it does, the medical pitch is marketing, not evidence — and the AI angle, conspicuously thin in the announcement, suggests Midjourney is leaning on its image-generation halo more than on any specific machine-learning capability the scanner actually requires.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
