Midjourney launches full-body ultrasound scanner and spa plan
Midjourney revealed the Midjourney Scanner, a 60-second ring ultrasound using 40 Butterfly modules and two petaflops of processing.
TL;DR
- 01Midjourney revealed the Midjourney Scanner, a 60-second ring ultrasound using 40 Butterfly modules and two petaflops of processing.
- 02Midjourney unveiled the Midjourney Scanner, an ultrasound-based full-body imaging system that captures vertical slices through a ring of sensors and produces 3D images in about 60 seconds.
- 03The device uses a ring of thousands of transducers, 40 Butterfly Ultrasound-on-Chip imaging modules per system, and two petaflops of processing power, the company said.
Midjourney unveiled the Midjourney Scanner, an ultrasound-based full-body imaging system that captures vertical slices through a ring of sensors and produces 3D images in about 60 seconds. The device uses a ring of thousands of transducers, 40 Butterfly Ultrasound-on-Chip imaging modules per system, and two petaflops of processing power, the company said.
What is the Midjourney Scanner and how does it work?
The Scanner records ultrasonic ripples passing through a person’s body as they descend on a platform into a shallow water pool, then reconstructs those signals into 3D images. A platform drops into water on rails through a ring of underwater sensors; the sensors send ultrasonic waves from many angles, the system captures the echoes, and software builds detailed images of muscle, fat, bone, and organs.
The hardware was developed in partnership with Butterfly Network and the company says each system includes 40 Butterfly Ultrasound-on-Chip modules. Midjourney’s demonstration combined the sensor ring with two petaflops of compute to create an overlayed segmentation view. The company says the scan process takes about 60 seconds and that about a dozen people have been scanned so far.
How developed is the device and what approvals will it need?
The system is a side project of Midjourney and remains in early stages, with the company acknowledging some medical applications will require FDA clearances. Midjourney described work focused first on “body composition maps” that do not require the same level of regulatory clearance as diagnostic imaging, while noting that diagnostic uses would need formal approvals.
Company job listings state an aim to “build and launch the world’s first full-body ultrasound CT scanner,” and Midjourney’s pitch includes a plan to let users create a library of scans that can be shared with doctors or AI health tools. The CEO, David Holz, said the company imagines getting fast internal views without magnets or radiation and suggested the scans could one day be comparable to MRI, saying the project "aims for image quality comparable to MRI in many ways." Holz also floated consumer-facing deployment: he hopes to place 10 scanners in a Midjourney Spa in San Francisco’s Union Square that the company plans to open before the end of 2027.
Why it matters
If the Scanner can scale beyond experimental demonstrations it would change the frequency and context of body imaging by removing the cost and logistical barriers of MRI and CT for non-emergency uses. Midjourney positions the device for routine monitoring, such as tracking how body composition responds to diet and exercise, and for consumer access via a spa concept. That shifts the conversation from isolated diagnostic imaging toward repeatable, preventative scans and large personal scan libraries shared with clinicians or software.
The medical value hinges on two things the company itself flagged: achieving reliable, clinically useful image quality and obtaining the necessary FDA clearances for diagnostic claims. Right now Midjourney is focusing on lower-barrier outputs like body composition maps while it builds the sensors, compute, and software pipeline.
What to watch
Look for regulatory milestones and clinical validation studies that test how the Scanner’s images compare to MRI for specific uses. Also watch for product deployments: Midjourney's plan to install 10 units in a Union Square spa before the end of 2027 would be an early, tangible rollout. Finally, monitor how the company defines data use and sharing as it moves from demo scans—about a dozen so far—to broader user collections and potential clinical sharing.
Written by The Brieftide · Source: The Verge
The Brieftide Daily · 06:00
Briefs like this one, in your inbox every morning.
Continue reading
More in Multimodal AIThinkDeception: Progressive RL framework for multimodal deception
ThinkDeception on arXiv uses MLLMs, a step-by-step multimodal Chain of Thought dataset and a four-tier progressive RL trainer for.
Visual-Seeker: visual-native multimodal search surpasses rivals
Zhengbo Zhang and 12 co-authors submitted Visual-Seeker on 13 Jun 2026.
Gemma 4 12B: unified, encoder-free multimodal model for laptops
Google DeepMind’s 12B model brings encoder-free vision and native audio to laptops, runs on 16GB memory and is released under Apache 2.0.
Hugging Face Spaces agents.md: chain image to 3D splats
An agent used two Hugging Face Spaces and their agents.md files to auto-generate images, reconstruct 3D Gaussian splats.