Related: AI Screens 6 Million Molecules to Find New Antibiotics for Drug Resistant Gonorrhea
Midjourney, the company best known for generating images from text prompts, has made an unexpected leap into medicine, unveiling a health division and a full body scanner built around sound rather than software. The new effort, called Midjourney Medical, centers on a device the company describes as an ultrasonic scanner, one that aims to produce a detailed three dimensional map of a person’s entire body in roughly 60 seconds using nothing but sound waves and water. It is a startling pivot for a firm whose entire reputation rests on pictures that exist only on a screen.
The physics behind the machine are genuinely novel. Rather than the handheld wand of a conventional ultrasound, the design calls for a ring made of around half a million tiny squares, each one the size of a fine grain of sand and each able to act as both a miniature speaker and a miniature microphone. As a person passes through the ring, those elements send sound into the body and listen to the echoes that bounce back, and software stitches the returning signals into a high resolution image. Because it relies on sound and water, the approach uses no ionizing radiation like a CT scan and no powerful magnets like an MRI.
The pitch is built on speed and resolution. Midjourney says the system is meant to reconstruct the body down to a fraction of a millimeter, a level of detail it compares to MRI, but at something close to a hundred times the speed once the technology matures. If that held up, it would turn a scan that today can mean a long wait, a narrow tube, and a large bill into something closer to walking through a doorway. The company has talked about an expansive future for the idea, floating a goal of tens of thousands of scanners performing on the order of a billion scans a month by the early 2030s.
The reality of the prototype is far more modest, and the company has been relatively upfront about that. The 60 second figure is an aspiration rather than a current capability, with early reporting suggesting a real scan today takes closer to 20 minutes, and the device has no regulatory clearance, which means it cannot be used to diagnose disease. Midjourney says it plans to launch by offering body composition maps, breakdowns of fat, muscle, and similar measures, rather than medical diagnoses, with a first location in San Francisco planned for late 2027. That framing keeps the product in the wellness and fitness lane for now, well short of replacing a hospital scan.
What makes the project more than a vanity experiment is the partner behind the hardware. The scanner is built on ultrasound on chip technology from Butterfly Network, a medical imaging company that shrinks ultrasound onto a semiconductor, under a co-development and licensing deal the two firms signed in late 2025 worth up to 74 million dollars. The current prototype uses 40 of Butterfly’s chips arranged in a ring, each packed with thousands of tiny transducers that together make up the roughly half a million sensors Midjourney describes. Butterfly’s chief executive, Joseph DeVivo, welcomed the announcement as a meaningful commercial opportunity for his company, and Butterfly’s stock rose on the news, a sign that an established player in medical imaging sees the effort as real rather than fanciful.
That does not make the path easy. Medical device veterans were quick to list the obstacles, from the years of research, regulatory review, and commercialization work any imaging device must clear to thornier questions about whether mass consumer scanning is even a good idea. Critics warn of overimaging, the risk that scanning healthy people surfaces harmless anomalies that trigger anxiety and unnecessary follow up, and of adding more work for already stretched radiologists asked to interpret the results. Building imaging hardware that satisfies regulators and clinicians is a long, capital intensive process that has humbled better funded efforts, and a company whose expertise is in generative AI is starting from outside the industry.
What makes the announcement worth watching anyway is the logic behind it. Midjourney’s core skill is reconstructing images from messy, incomplete information, which is precisely the computational problem at the heart of turning scattered sound echoes into a clean picture of the body. Seen that way, an image company moving into imaging hardware is less random than it first appears, even if the engineering and regulatory road ahead is daunting. The company is betting that the part it is best at, the math of forming an image, can be paired with new sensors to do something useful in the physical world.
For now the scanner is a bold prototype with a long list of unproven claims and a 2027 target that leaves plenty of time for the picture to change. If it works, it would make detailed body imaging dramatically faster and cheaper, a genuinely meaningful shift. If it does not, it will be a reminder that medicine is a far harder arena than the screen, and that turning sound into a trustworthy image of the human body is a problem no amount of prompt engineering can shortcut.
Related on Entrelligence: AI screening 6 million molecules for new antibiotics, and Neuralink’s through-the-dura brain implant.
Your First 10 AI Skills
10 practical AI skills, copy-paste prompts and a 7-day plan to start using AI with confidence.
Download the guide →