Midjourney goes from generating cat images to full-body ultrasound scans

Midjourney’s founder David Holz took the stage recently to demonstrate a leap that feels almost science‑fiction. In an interview cited by The Verge, he rolled out the Midjourney Scanner, a ring of ultrasound sensors that capture vertical slices of the entire body, mapping muscle, fat, bone and organs in a single pass.

The device is billed as a full‑body ultrasound that can deliver image quality “comparable to MRI in many ways.” Holz says the goal is to make scans as easy as a yearly check‑up or even a daily glimpse into your own physiology, a far cry from the AI‑generated cat memes that put Midjourney on the map.

What makes this move interesting is the convergence of generative AI and medical hardware. Midjourney’s strength has always been in turning text prompts into vivid visuals; now it’s applying that visual understanding to real‑world data. The scanner’s raw images could be fed straight into the company’s existing diffusion models for automatic tissue labeling, anomaly detection or personalized health reports.

From a broader tech perspective, the announcement signals an emerging trend where AI startups expand beyond software into hardware that captures the data their models need. Companies like Runway and Stability AI have hinted at similar ambitions, but Midjourney’s first tangible product puts a spotlight on the regulatory and safety hurdles that accompany medical devices.

If the scanner lives up to its promise, it could democratize high‑resolution imaging. An affordable, portable ultrasound that rivals MRI could shift some diagnostics from hospitals to clinics, gyms, or even home wellness centers. That said, the path to FDA clearance is steep, and privacy concerns around continuous body monitoring will need clear policies.

For entrepreneurs, the Midjourney Scanner opens a new frontier for AI‑driven health services. Imagine a subscription that pairs daily scans with a cloud‑based AI that flags subtle changes in tissue composition, prompting early intervention before a condition becomes serious.

What this means for you: if you already use AI tools for content creation, consider extending them to health data analysis. A practical experiment is to upload a sample ultrasound image to a vision model like GPT‑4V and ask it to identify major tissue types. Here’s a quick prompt to try:

"Analyze this ultrasound slice and list any visible muscle, fat, bone and organ structures, noting any irregularities that could indicate a health issue."

Running that workflow through an AI assistant could give you a taste of how visual AI can augment medical imaging, a skill set that may become increasingly valuable as hardware like the Midjourney Scanner enters the market.