Waves of Innovation: The Ultrasonic CT Revolution

AI DiagnosticsAI & Machine LearningAI Therapy

Imagine a whole-body scan that uses sound waves instead of radiation—Midjourney Medical is making that a reality.

TL;DR
• Midjourney Medical launched an ultrasonic CT scanner in 2026, still awaiting FDA clearance.
• FDA has cleared 1,451 AI medical devices, mostly in radiology, with rapid funding growth.
• Success hinges on narrow focus, real-world validation, and workflow integration, not hype.

Summary
Midjourney Medical unveiled a whole‑body ultrasound scanner in June 2026, featuring 358,000 transducers, 40 GB per slice data processing across 21 servers, and 60‑second 3D imaging, yet it lacks FDA diagnostic clearance and is positioned initially as a wellness body‑composition service. The broader AI‑enabled medical device market grew to 1,451 clearances by end‑2025, with radiology accounting for 76 % and major players such as Aidoc, Viz.ai, GE HealthCare, Siemens Healthineers, Tempus AI, and PathAI demonstrating extensive hospital adoption and multi‑site validation.
• Prioritize narrow, outcome‑driven use cases and secure prospective multicenter studies before scaling AI tools.
• Build data pipelines that ingest diverse, multi‑vendor imaging streams to mitigate the 24 % specificity loss observed when models leave their training environment.
• Embed continuous monitoring and bias audits to detect shortcut learning and demographic performance gaps early in deployment.

Midjourney’s partnership with Butterfly Network includes a $15 M upfront fee and up to $74 M in milestone payments, but its first clinic—a 25,000‑sq‑ft spa in San Francisco—won’t open until late 2027, and the company projects 50,000 scanners and one billion scans per month by 2031 without published clinical validation. In contrast, Aidoc’s foundation model cleared in January 2026 covers 14 CT conditions with 97 % sensitivity and 98 % specificity, processing 60 million cases annually across roughly 2,000 hospitals. Viz.ai’s stroke detection algorithm reduces treatment time by 31 minutes and 90‑day disability by 40 % at over 1,700 hospitals, illustrating the impact of tightly integrated, time‑critical AI.

Regulatory pathways remain dominated by 510(k) clearances (95‑97 % of approvals), limiting the depth of clinical evidence required. Companies achieving De Novo or Breakthrough Device Designation, such as PathAI’s AISight Dx and Tempus AI’s Paige acquisition, demonstrate the value of deeper validation and strategic M&A to assemble large, diverse datasets for foundation models. Nvidia’s Clara platform contributes open‑source models for explainable chest X‑ray interpretation and synthetic data generation, supporting the need for adaptable, extensible AI ecosystems.

The sector’s challenges include shortcut learning, demographic bias, and HIPAA‑compliant data handling, all of which demand robust governance and legal review. Success in 2026 correlates with modular hardware, scalable cloud‑based processing, and clear regulatory pathways, suggesting that future investments should focus on flexible architectures, continuous performance monitoring, and partnerships that accelerate data diversity and model robustness.

Will Midjourney’s bold hardware bridge the gap between body maps and diagnostic certainty?