Midjourney Medical and the Imaging Gap Nobody Built

The Debate Is Asking the Wrong Question

The clinical community's response to Midjourney Medical's announcement has centered on one comparison: is this as good as MRI? That is the wrong question.

The right comparison is between the Midjourney Medical scanner and the gap that exists before a patient ever reaches advanced imaging, and what happens to the people who never do. Not every clinical question requires advanced imaging, and advanced imaging is not without risk.

Midjourney Medical's scanner is technically distinct from standard point-of-care ultrasound. Using sound waves and water, with no radiation and no magnetic field, it performs a full-body tomographic scan in 60 seconds. Built on Butterfly Network's Ultrasound-on-Chip platform, with 40 chip modules per prototype system, it is designed to generate 3D maps of the body at resolution comparable to MRI at a fraction of the cost and time. Butterfly Network supplies the semiconductor and software infrastructure under a five-year co-development agreement worth up to $74 million. Midjourney Medical owns the device, the commercial model, and the regulatory pathway.

The clinical concerns that some have raised about false positives and unnecessary downstream care are legitimate and deserve honest engagement. It also deserves honest and consistent application to the system already in place.

What Advanced Imaging Does Not Do

Advanced imaging is expensive, intermittent, and inaccessible for large portions of the population. It is ordered in response to symptoms, referrals, or screening protocols that themselves carry significant gaps. It is not designed for longitudinal surveillance. It is not designed for the asymptomatic patient with family history and subclinical risk factors who has no clinical trigger for a formal diagnostic workup.

It also carries risk. Ionizing radiation from CT imaging is a documented carcinogen at population scale. The evidence on lifetime cumulative CT exposure and malignancy risk is not ambiguous. What is unknown is what any individual patient's lifetime exposure actually is, because the infrastructure to track it does not exist in any functional sense.

Radiation dose does not follow the patient across health systems. In many cases it does not follow the patient within a single health system. DICOM images, the standard format for medical imaging data, are not meaningfully integrated into most EHRs. Patients accumulate imaging history across institutions, modalities, and decades, and no clinician treating them today has reliable access to that longitudinal record.

The structural barriers to image sharing are well understood. They have simply never been prioritized:

• File size and storage. DICOM images are orders of magnitude larger than structured clinical data, creating infrastructure demands that most HIEs were not built to absorb.
• Interoperability architecture. HIEs run on FHIR and CCDA per USCDI standards. Imaging systems live in DICOM PACS and VNA silos, often on-premises, outside the exchange infrastructure entirely.
• Financial disincentives. Some organizations benefit from repeat imaging. The incentive to share prior studies is not universal, and in some cases it runs directly against revenue.
• Privacy and compliance complexity. Patient metadata embedded in DICOM images exceeds the compliance surface of standard clinical data exchange.
• Workflow friction. Most PACS and VNA systems were designed for local network access, not seamless cross-institutional exchange. Patients are still being told to bring images on a disc.

None of these barriers is technically intractable. Efforts like The Sequoia Project, CommonWell Health Alliance, and TEFCA have made meaningful interoperability progress on clinical documents and data. Images remain the missing chapter. Interoperability without images is like a medical record without the scans. You get the words, but you lose the clinical story that only the image can tell.

This is not a technology limitation. It is a set of commercial choices that have never been reversed at scale. The incumbent imaging industry built its model around capital equipment, service contracts, and radiologist relationships. Interoperability was not a priority because fragmentation served the installed base. New studies get ordered partly because prior studies are inaccessible, reports are not trusted without the images, and ordering a new study is the path of least resistance.

Why Does Outpatient POCUS Still Lag?

Outpatient POCUS adoption has lagged because the value case requires a payer and HEOR argument that the incumbent imaging industry has never built. Point-of-care ultrasound grew rapidly in acute care for a structurally simple reason: the value case was self-evident and internal to a single institution. Bedside echo in the ICU replaces a formal echo consult and produces the answer faster. A hospital CFO or CMO could see that value without a payer coverage argument or a health economics brief.

The value case in outpatient requires a different argument, one about reducing downstream utilization, improving access in underserved settings, and lowering total cost of care across an episode. That is a payer and HEOR argument. The large traditional imaging OEMs have never built that capability. Philips, Siemens Healthineers, and GE HealthCare have no meaningful market access function oriented around payer value. Their commercial model was built around capital equipment cycles, service contracts, and radiologist relationships. Payers paid because hospitals bought the machines and radiologists billed for the reads. Market access was someone else's problem.

Butterfly Network is a different animal. Their foundational bet was architectural: put the ultrasound transducer on a single semiconductor chip, make the device software-upgradable and phone-connected, and build toward a platform rather than a product. That orientation toward integration over isolation is precisely what separates them from the incumbent OEM model. They have been creative in advancing the clinical case for point-of-care imaging in ways the large capital equipment companies never pursued. The challenge the entire category faces is not a technology design problem. It is a market access and HEOR infrastructure problem that no player at commercial scale has yet solved for the outpatient setting.

The consequence is that POCUS in outpatient settings has remained underleveraged relative to its clinical potential, not because the clinical evidence is weak, but because the reimbursement infrastructure has never been built to support it at scale.

AI in imaging has followed the same pattern among the traditional OEMs. Features are abundant. Products with independent coverage pathways are not. AI capabilities have been bundled into capital equipment sales because that was the only commercial motion available to companies organized around hardware revenue. The result is that imaging AI is largely uncoded, unreimbursed, and invisible to payers as a standalone value driver. It cannot be evaluated independently because it was never positioned independently.

What Is the Midjourney Medical Scanner?

Midjourney Medical is a new division of Midjourney, the AI image generation company, focused on whole-body health imaging. Its first product, the Midjourney Scanner, performs a full-body ultrasonic CT scan in 60 seconds. The user stands on a platform that descends into water, passing through a ring of approximately 500,000 ultrasonic transducers, each acting as both a transmitter and receiver. The system collects terabytes of data per second and reconstructs a detailed 3D map of the body.

The device is built on Butterfly Network's Ultrasound-on-Chip semiconductor platform. The current prototype uses 40 Butterfly chip modules. Future generations are expected to use substantially more, with image quality described as substantially improved by generation three. Midjourney Medical plans to launch its first commercial site, a wellness facility in San Francisco, by end of 2027, with a target of 50,000 scanners worldwide by 2031.

The commercial strategy is wellness-first, which is the correct regulatory sequencing. Regulatory clearance for broader diagnostic use will follow as clinical evidence accumulates. Critics will focus on the wellness framing. The delivery channel does not determine the clinical question. The question is whether longitudinal, radiation-free, whole-body imaging at population scale creates a preventive surveillance layer the current system cannot offer. That question is serious regardless of the address.

The Pathway Nobody Built

What Midjourney Medical is proposing is not a replacement for advanced imaging. It is a radiation-free, connected, AI-enabled surveillance baseline that creates a documented pathway to advanced imaging when clinically warranted. The patient who gets longitudinal low-cost ultrasound surveillance and catches a meaningful finding early is not the patient who gets an unnecessary MRI. They are the patient who gets the right MRI at the right time with clinical context already established.

That use case is largely unserved by the current system. No incumbent imaging company has built a patient-centered pathway across modalities. The focus has been on products and installed base, not on how imaging data follows a patient across a lifetime of care.

To realize that value, Midjourney Medical will need what the incumbent imaging industry never built:

• A market access function with fluency in payer coverage logic and reimbursement pathways
• A reimbursement strategy that distinguishes wellness positioning from the covered benefit pathway and maps the evidence requirements for the transition
• HEOR demonstrating that longitudinal low-cost ultrasound surveillance reduces downstream high-acuity utilization and total cost of care
• A data architecture that allows imaging to follow the patient across care settings rather than staying siloed at the point of acquisition

None of that is in the current announcement. All of it is required for this to become more than a consumer subscription. The commercial road from wellness facility to covered benefit is long and has no established map. But the clinical and economic rationale for building it is real, and the white space is largely uncontested.

The Democratization Argument

Advanced imaging is structurally inaccessible for large portions of the population. Rural communities, underserved urban populations, and patients without robust insurance coverage do not get the same access to diagnostic imaging as patients with resources and proximity to academic medical centers. That disparity has downstream consequences for disease detection, treatment timing, and outcomes.

Longitudinal, low-cost, zero-radiation imaging that follows the patient, builds a baseline over time, and flags meaningful change early is not a consumer novelty. It is what a preventive, patient-centered system should have been building toward all along. The growth of Direct Primary Care, functional medicine, and direct-to-consumer pharmaceutical models including GLP-1s are not isolated trends. They are market responses to a system that has failed to serve patients who can afford to route around it.

Direct Primary Care is the clearest signal. An 83% surge in practices opting into hybrid and direct models, documented in a 2025 Health Affairs study, reflects a market-level judgment that administrative friction and reimbursement uncertainty have become structurally prohibitive. When clinicians with full panels of insured patients choose to exit the insurance system, that is not a lifestyle decision. It is a signal about value chain failure. The patients who cannot afford to route around the system are the ones the current imaging infrastructure was never designed to reach.

The imaging access disparity is not hypothetical. Post-2021 USPSTF guideline changes tripled lung cancer screening participation but exposed persistent racial and ethnic gaps, with representation among Black and White patients declining relative to the newly enrolled population. Guideline changes alone do not close access gaps. Active outreach and the removal of structural barriers do. Radiation-free, lower-cost imaging at scale is a structural intervention, not a premium amenity.

The question is not whether Midjourney Medical's scanner is as good as MRI. The question is whether it serves the people the current system was never designed to reach. Healthcare is about people caring for other people in the fullest, most informed, and safest manner possible. A system that cannot tell a clinician what a patient has already been exposed to, or what imaging they already have, is not meeting that standard.

The loudest criticism of this announcement is coming from the industry that had every opportunity to build this first and chose not to. That is worth noting.

Frequently Asked Questions  ·  Erik's Hot Take