Computer vision for early tumor detection works in at least one large randomized trial (Sweden’s MASAI, 2026).
Regulators now require more lifecycle control, but most cleared devices still reach the market without a prospective clinical study. Only tumor and cancer detection cases are treated as case studies. Sources: Full paper, PMC · DOI 10.1001/jamadermatol. 2019.1735 Case 3.
Key takeaways
- Sources: RSNA press release · DOI 10.1148/radiol.222176 Case 4.
- Recalls clustered among devices with no reported clinical validation.
- These cover all AI devices, not only tumor detection.
Summary: The counter-evidence: where it works An honest case study needs the other side.
Read the full article: https://www.ainvasion.com/ai-tumor-detection-failures/
Top comments (0)