Midjourney has decided to step into one of the most delicate areas of technology: medical imaging. The company, known for its AI-powered image generation models, has introduced Midjourney Medical along with a prototype for an ultrasound-based body scanner that promises to produce a 3D body map in about 60 seconds, without ionizing radiation and without the magnetic fields used in MRI.
The announcement is eye-catching both for what it proposes and for who is behind it. Midjourney does not come from the medical hospital sector, radiology, or medical devices. It originates from visual creation using AI. While this background explains part of the enthusiasm, it also requires a more rigorous look at the project. In medicine, it’s not enough to generate visually stunning images: proof of correctness, reproducibility, clinical utility, and safety for users are essential.
The company envisions an ambitious goal: to make body scanning a regular routine, almost as simple as visiting a spa. The first Midjourney Spa is planned for San Francisco in 2027, combining scanners, swimming pools, saunas, and cold baths. The aim is not to start with full medical diagnoses but with body composition maps and visual tracking over time.
A sci-fi aesthetic ultrasound scanner
Midjourney describes a functioning process different from conventional ultrasound. The person enters shallow water, steps onto a platform, and slowly descends through a ring of ultrasonic sensors. This ring emits sound waves from multiple angles and records how they deform as they pass through tissues of varying density and rigidity.
Then, the system reconstructs this data into 3D images and applies AI segmentation to identify internal structures. Midjourney calls it “Ultrasonic CT,” although it does not use X-rays like a CT scan. The concept is closer to immersion ultrasound tomography, with significantly more data capture and processing than a manual ultrasound exam.
| Technology | How it works | Advantage | Main limitation |
|---|---|---|---|
| CT Scan | X-ray from multiple angles | Speed and high resolution | Ionizing radiation |
| Magnetic Resonance | Magnetic fields and radiofrequency | Great soft tissue contrast | Cost, time, restrictions |
| Traditional Ultrasound | Handheld probe with ultrasound | Safe, affordable, widely available | Limited field and operator dependence |
| Midjourney Scanner | Water ring ultrasound with 3D reconstruction | Wide scan area and radiation-free | Lack of clinical validation |
Water-based design makes technical sense because it improves acoustic coupling. Conventional ultrasounds use gel for better transmission between probe and skin, but in this case, water acts as a continuous medium around the body, allowing sound waves to be sent from multiple points simultaneously.
According to Midjourney, the result would be a volumetric map of the body with sub-millimeter detail and a look similar to MRI. While this comparison is compelling, it’s also risky. Generating images that look similar visually is one thing; achieving the same clinical utility is another.
Butterfly Network provides the actual ultrasound foundation
The most concrete part of the announcement relates to the collaboration with Butterfly Network. The U.S. company confirmed that the current prototype uses 40 Butterfly Ultrasound-on-Chip modules per system, integrated under a co-development agreement. Butterfly is renowned for its semiconductor-based ultrasound devices, replacing parts of traditional technology with chips containing thousands of transducers.
This detail reduces some of the mystery. Midjourney has not invented a new medical imaging physics from scratch. Instead, it combines ultrasound-on-chip modules, water ring capture, computational reconstruction, AI, and user experience design. The innovation lies in system integration, scalability, and the overall capture workflow.
| System component | Value contributed by |
| Ultrasound-on-chip | Butterfly Network |
| Scanner design | Midjourney Medical |
| Computational reconstruction | Midjourney and its computing infrastructure |
| AI segmentation | Image analysis models |
| Spa-like experience | Midjourney |
| Medical validation | Pending studies and regulation |
Butterfly also benefits from this operation. Its ultrasound-on-chip technology would extend from portable devices to a comprehensive body imaging platform. If successful, it could open a new category for its modules. Even if it doesn’t, it underscores that semiconductor-based ultrasound can go far beyond traditional clinical probes.
The big challenge: ultrasound doesn’t see everything equally
Ultrasound offers clear advantages: it doesn’t use ionizing radiation, can be relatively inexpensive, is fast, and has decades of clinical use. However, it also faces physical limitations. Sound waves do not penetrate all tissues equally. Bone, air, and certain interfaces can block, reflect, or degrade signals.
This is crucial for understanding why comparing the system to MRI might be misleading. MRI provides valuable information for brain, joints, abdomen, spinal cord, heart, or soft tissues. Ultrasound excels in many applications but cannot simply replace all imaging modalities without limitations.
| Clinical question | Why it matters |
| Which organs can be reliably reconstructed? | Defines initial real-world applications |
| How does it handle bony areas? | Limits cranial, spinal, and similar regions |
| What about air and intestinal gas? | Can affect abdomen and thorax imaging |
| What about false positives? | Prevents unnecessary alarms |
| What about false negatives? | Prevents false reassurance |
| How does it compare with MRI, CT, or DEXA? | Determines clinical value |
| Who interprets results? | Defines clinical responsibility |
This does not invalidate the project; it simply places it within realistic boundaries. Midjourney could develop a useful tool for body composition, muscle and fat tracking, surface organ monitoring, or longitudinal changes—without replacing MRI. Its utility will depend on indications, not slogans.
From hospital to spa: a bold yet problematic idea
Planning to present the scanner within a spa setting is one of the project’s most intriguing aspects. Midjourney aims to make body scans feel less like medical exams and more like frequent, pleasant, and comfortable experiences. From an adoption perspective, it makes sense, as many people avoid medical tests due to time, cost, fear, or discomfort.
However, this also sparks debate. Moving body imaging into wellness environments could bring preventive health closer to more people, but it risks blurring the lines between wellness, self-care, and diagnosis. If users receive internal body maps without proper medical oversight, misinterpretation, anxiety, and hasty decisions could result.
| Potential benefit | Associated risk |
| More frequent scans | Overinterpretation of findings |
| Less intimidating experience | Confusion between wellness and medicine |
| Long-term body tracking | Anxiety over minor changes | More accessible | If costly, could deepen inequalities |
| Data shareable with doctors | Privacy and consent issues |
| Prevention with visuals | False positive risks |
Preventive medicine isn’t just about more measurements—it’s about better measurement, informed interpretation, and acting upon validated evidence. Midjourney will need to demonstrate that its spa-like experience doesn’t turn the body into a constant source of concern.
The FDA as the ultimate gatekeeper
Midjourney acknowledges that diagnostic capabilities will require regulatory approval. They say they’ll start with body composition maps and submit results to the FDA to expand functionalities. This is critical: showing a body representation is not the same as claiming a device can detect diseases.
The FDA assesses AI medical devices based on safety, effectiveness, risk management, training data, validation, performance, bias, cybersecurity, and post-market monitoring. Each clinical application must be clearly defined. A system might be valid for estimating body composition but not approved for cancer detection, vascular disease, or musculoskeletal injuries.
| Possible phases | What must be demonstrated |
| Research | Basic safety and signal quality |
| Body composition | Accuracy compared to existing methods |
| Longitudinal tracking | Reproducibility over time |
| Specific clinical use | Sensitivity, specificity, usefulness |
| Diagnosis | Regulatory approval for specific indications |
| International scaling | Compliance with local regs and data management |
This process could take years. Validation must include diverse populations, different body types, ages, clinical conditions, and use scenarios. A system that works in a handful of individuals isn’t ready for widespread healthcare deployment.
Body data: Midjourney’s most sensitive asset
Privacy considerations are enormous. A complete record of body scans could be far more sensitive than a photo gallery, browsing history, or activity log. It could reveal anatomical changes, body composition shifts, anomalies, habits, treatment effects, or disease progression.
Midjourney must clarify how it will store, encrypt, process, and share this data. It should also specify if scans can be used to train models, under what consent, and through what mechanisms of deletion or transferability.
| Privacy question | Why it matters |
| Where are scans stored? | Sensitive health data | Are they used for training? | User consent and control |
| Can spa staff access? | Operational risks |
| How are they shared with doctors? | Interoperability and security |
| What if there’s a breach? | High personal impact |
| Can users delete their history? | Control over data rights |
| Will there be secondary commercial use? | Market trust |
This aspect may end up being as important as image quality. Acceptance of healthcare technology depends not just on functionality but also on trust in those operating it.
A roadmap more typical of an AI startup than traditional medtech
Midjourney envisions a 12-month phase to refine algorithms and hardware, an experimental spa by late 2027, a third-generation device with custom silicon in 2028, and a goal of 50,000 scanners in 2031 capable of 1 billion scans per month.
This is an extremely aggressive roadmap for the medical field. AI software can iterate rapidly, but medical hardware cycles are slower: manufacturing, quality, certification, calibration, maintenance, insurance, regulation, training, legal liability, and clinical evidence all extend timelines.
| Announced goal | Prudent reading |
| 60-second body scan | Technical target pending broad validation |
| Spa in San Francisco in 2027 | First real-world operational lab |
| Gen3 with custom silicon in 2028 | Major engineering leap |
| 50,000 scanners by 2031 | Very demanding industrial scale |
| 1 billion scans per month | Long-term vision, not proven capacity |
| Future diagnostic use | Dependent on FDA approval and clinical studies |
Speed can be an advantage for rapid prototyping but risky if products outpace evidence. Healthcare heavily penalizes overpromising solutions.
The opportunity: making visible what is now seldom measured
Despite caution, the project has genuine potential. Most people only access medical imaging when symptoms or suspicion arise. Between a home scale and MRI, there’s room for more informative, safe, and well-communicated body tracking tools.
A quick scanner could monitor body composition, muscle changes, fat distribution, or physical evolution during training, aging, recovery, or treatment—producing valuable data for research when properly managed with consent and security.
The key is whether Midjourney can turn this vision into a reliable, affordable, and clinically meaningful device. To do so, it will need collaboration with clinicians, radiologists, regulators, bioengineers, privacy experts, and health systems. The spa aesthetic may draw attention, but it doesn’t replace clinical work.
Trust as the real test
Midjourney Medical is a bold venture because it enters a territory where AI companies can’t get by with mere demos. A mistake in a creative image can be bothersome. A mistake in a body image can influence health decisions.
The company has an advantage: it knows how to turn complex technology into easily understandable experiences. Its challenge will be doing so without oversimplifying the complex realities of medicine. If it presents this as a body tracking tool rather than promising to replace MRI, there’s room to learn. But if it claims “to substitute MRI,” scrutiny will intensify.
This project is worth watching because it points to a likely future: AI moving from screens into sensors, devices, medical imaging, and real-world spaces. Midjourney may have found a new frontier. But in healthcare, crossing that frontier depends on data, validation, and trust—not promises.
Frequently Asked Questions
What is Midjourney Medical?
It’s the new division of Midjourney focused on body imaging and health technologies, starting with a full-body ultrasound-based scanner.
Is the Midjourney scanner approved for diagnosis?
No. Currently, it’s presented as a research and body tracking project. Diagnostic capabilities will require regulatory approval.
What does Butterfly Network contribute?
Butterfly supplies Ultrasound-on-Chip modules. The current prototype incorporates 40 modules per system under a co-development agreement.
Can it replace MRI?
There isn’t enough public evidence yet to confirm that. Ultrasound has advantages but also physical limits, especially with bone and air. Each clinical application must be validated separately.
Why does Midjourney talk about a spa?
They want to make body scanning a frequent and comfortable experience within wellness environments. While this may ease adoption, it also requires clear communication to avoid confusing wellness with medical diagnosis.
Source: Noticias.ai and Midjourney
