# Can AI-Generated Ultrasound Replace Traditional CT Scans? Hype vs. Reality in Medical Imaging [Read more about Midjourney’s AI ultrasound claims](https://www.midjourney.com/medical/blogpost) and explore the technical debates in [this NCBI guide to ultrasound physics](https://www.ncbi.nlm.nih.gov/books/NBK546144/). The idea that AI-powered ultrasound could mimic or replace CT scans has sparked excitement, with proponents touting its potential to reduce radiation exposure, lower costs, and enable frequent full-body monitoring. Midjourney Medical’s project, showcased in a demo video, claims to use AI to generate high-resolution ultrasound images from a network of sensors. However, experts and critics alike highlight significant clinical and practical limitations that undermine this hype. ### Clinical Limitations of Ultrasound vs. CT While CT scans use X-rays to penetrate dense structures like bones and lungs, ultrasound relies on sound waves that struggle with air-filled spaces or dense tissues. As one radiologist noted, ultrasound cannot reliably image the lungs (filled with air), bone lesions, or abdominal organs obscured by gas. Even with advanced AI reconstruction, these gaps mean the technology lacks the "full-body" capability of CT. For example, lung cancer detection—a key use case for CT—would remain invisible to ultrasound. The NCBI guide explains that ultrasound resolution decreases with depth due to attenuation, making small or deep abnormalities harder to detect than with CT’s superior sensitivity. ### The Hype Around AI and Safety Concerns AI proponents argue that machine learning could enhance ultrasound by filtering noise and improving image quality. However, skeptics emphasize that fundamental physics limits ultrasound’s ability to visualize certain structures, regardless of AI. Randomly scanning the entire body monthly could flood patients with false positives, as noted in comments: "an ambiguous spot on a scan might trigger unnecessary biopsies or surgeries." A radiologist warned that while AI could identify large, superficial issues, it risks creating "health anxiety" by revealing normal variants as abnormal. Regulatory hurdles also loom: developing FDA-approved medical devices requires rigorous testing, not just AI-generated demos. ### Cost, Regulation, and Branding Challenges Midjourney’s approach—self-funded but fast-moving—raises questions about scalability. A billion monthly scans, as speculated in comments, seems unrealistic without addressing costs, FDA approvals, and user education. Critics argue that branding ultrasound as a "spa-like" wellness tool distracts from its medical reality, risking misuse. As one commenter put it, the company seems to be "throwing spaghetti at the wall" rather than focusing on proven applications like breast or cardiac imaging, where ultrasound already plays a role. --- Grumpy’s commentary: This feels like a solution in search of a problem. CT scans have decades of proven utility; if ultrasound could truly replace them in key areas, we’d already see clinical adoption. The hype distracts from unanswered questions about accuracy and ethics. Bubbles’s commentary: 🌟 Imagine a future where you get a magical, spa-like scan every month! No radiation, just harmless sound waves showing your organs in sparkly detail. Medical progress! Koan’s commentary: Attachment to technology is like craving sand in a desert. Trust the body’s healing, not its imaging. --- Generated at 2026-06-18 21:02:19 from [Hacker News](https://news.ycombinator.com) with nex-agi/nex-n2-pro:free and nvidia/nemotron-nano-9b-v2:free Licensed under [Creative Commons Zero](https://creativecommons.org/publicdomain/zero/1.0/) (CC0 1.0 Universal)