Videa Dental AI

Videa Dental AI is a cloud-based software device that uses artificial intelligence to automatically detect and localize dental findings in intraoral radiographs (bitewing, periapical, and panoramic images). It identifies suspected dental findings (caries, attrition, broken teeth, etc.), historical treatments (crowns, fillings, implants, etc.), and normal anatomy features in patients aged 3 years and older. The device is intended as an adjunct tool for trained dental professionals and should not replace a dentist's clinical judgment.

Videa Dental AI adds two new features to the predicate Videa Dental Assist: (1) segmentation output view (isocontour) as an alternative to bounding boxes for caries and calculus indications, and (2) a second operating point toggle allowing users to select between high sensitivity versus high specificity modes for caries and periapical radiolucency. Both devices use supervised deep learning on X-ray radiographs and display results in a dental image viewer; otherwise they share the same technological foundation and AI architecture.

ISO 14971:2019 (Risk Management), AAMI CR34971:2022 (AI/ML guidance), IEC 62304 Edition 1.1 (Software lifecycle), FDA Content of Premarket Submissions for Device Software Functions (June 2023), and Good Machine Learning Practice for Medical Device Development (October 2021).

Videa Dental AI is substantially equivalent to Videa Dental Assist because both devices have identical indications for use, the same intended patient population and users, and use the same underlying AI algorithms with no retraining for lesion detection. The new segmentation view and dual operating point features do not raise different safety or effectiveness questions: clinical testing showed no statistically significant performance difference between segmentation and bounding box views across 8 suspect dental finding indications, and both operating points demonstrated statistically significant clinician benefit. Performance metrics (bench DICE scores, clinical AFROC FOM improvements, specificity values) are comparable between the two devices, and both met identical acceptance criteria with no adverse events observed.