Sonio Detect (v3)

Sonio Detect v3 is a software-as-a-service (SaaS) solution that analyzes fetal ultrasound images and clips using machine learning to automatically detect views, identify anatomical structures within those views, and verify quality criteria and characteristics. It is intended for use as a concurrent reading aid during acquisition and interpretation of fetal ultrasound examinations across all three trimesters (11 to 41 weeks gestation), assisting healthcare professionals (sonographers, OB/GYN specialists, and fetal surgeons) in ensuring examination completeness and protocol compliance.

Sonio Detect v3 differs from predicate Sonio Detect v2 in two main ways: (1) v3 automatically localizes views and anatomical structures by outputting bounding boxes, whereas v2 does not; and (2) v3 uses Artificial Intelligence alone, while v2 also included lecture of biometrics and colorimetry for 3D and Doppler. Both are cloud-based and stand-alone software compatible with ultrasound systems from GE Medical, Samsung, Canon, and Philips.

Not stated in this summary.

Sonio Detect v3 is substantially equivalent to predicate Sonio Detect v2 (K240406) because both devices share the same intended users, intended use, and clinical applications (concurrent reading aid for fetal ultrasound). The addition of bounding box localization and simplified algorithm methodology do not raise new safety or effectiveness questions. Verification, validation, and bench testing on 22,496 independent fetal ultrasound images demonstrated strong performance (sensitivity 0.81–1.0, specificity 0.80–1.0 across all functions), with validation across ultrasound manufacturers, BMI, maternal age, image quality, geography, gestational age, and race/ethnicity subgroups.