On August 26, 2026, the Danish medtech company ROPCA announced FDA 510(k) clearance for two linked products: ARTHUR, the Arthritis Ultrasound Robot, and DIANA, the AI software that reads what ARTHUR captures. ARTHUR performs standardised musculoskeletal ultrasound of the finger, hand, and wrist joints under healthcare-professional supervision; DIANA analyses the resulting images and supports automated report generation. ROPCA describes the pair as an end-to-end workflow combining consistent image acquisition with AI-assisted analysis.
This is a different shape of clearance from most AI device authorisations. The large majority of FDA-cleared AI in medicine is software reading images that a human sonographer or radiographer produced. Here the acquisition itself is automated: a robot positions the probe and takes the scan, and the AI reads it. Removing the human operator from acquisition also removes the largest source of variability in musculoskeletal ultrasound, which has historically been operator dependence severe enough to limit the modality’s use in rheumatology.
The clinical base behind the submission is European rather than American. ARTHUR has been CE-marked since 2022 and ROPCA cites more than 5,400 scans and 83,500 joints scanned across six European countries, plus Mayo Clinic validation work for research applications. The commercial argument in the announcement is workforce arithmetic: 72 percent of US counties have no rheumatologist, and ROPCA points to projections of a shortfall of roughly 4,729 specialists against demand by 2030.
For health system leaders, this is the case worth watching, because it tests whether automation of the procedure, not just interpretation of it, can pass a regulator and then survive reimbursement and workflow. If it does, the scarce-specialist bottleneck in a whole discipline moves from “who can scan” to “who signs off.”