Overjet Caries Assist-Pediatric

Overjet Caries Assist-Pediatric (OCA-Ped) is an AI-powered software utilizing a machine learning model to analyze dental radiographs, functioning as a computer-assisted detection tool to help clinicians identify and segment carious lesions on bitewing and periapical images in children aged 4-11. Clinical studies demonstrate that OCA-Ped improves dentists' caries detection sensitivity, serving as an adjunctive diagnostic aid rather than a replacement for professional clinical judgment.

Overjet Caries Assist-Pediatric (OCA-Ped) is a software-only medical device designed as a radiological, automated, concurrent-read, computer-assisted detection (CADe) tool. Its primary purpose is to assist dental clinicians in detecting and segmenting dental caries on bitewing and periapical radiographs for pediatric patients aged 4-11 years, who may have primary, permanent, or mixed dentition. The system operates using a three-layer architecture: a Network Layer to ingest radiographic images, a Decision Layer where a machine learning model analyzes these images to predict carious lesions, and a Presentation Layer that displays these predictions as adjustable polygon annotations on a dashboard. OCA-Ped is intended as an adjunctive tool to provide additional diagnostic information and is not meant to replace a dentist's comprehensive review or clinical judgment, which incorporates patient history and in vivo clinical assessments. The performance of OCA-Ped was evaluated through multi-reader multi-case (MRMC) studies and standalone testing. In the MRMC study involving 20 dentists and 636 images, clinicians assisted by OCA-Ped demonstrated a statistically significant 7.5% improvement in caries detection, measured by the area under the weighted Alternative Free-Response Receiver Operating Characteristic (wAFROC) curve, compared to unassisted reading. This was accompanied by an 11.8% increase in average tooth-level sensitivity, although there was a slight 1.1% decrease in average tooth-level specificity. Standalone performance testing on 1190 images showed a tooth-level sensitivity of 83.9%, a specificity of 97.5%, and a mean Dice coefficient of 79.0% for lesion segmentation, with performance varying slightly across different X-ray sensor brands. These results indicate the device enhances clinicians' ability to identify caries in the specified pediatric population.

AI Regulatory Concerns

Powered by

Jev reads the full FDA letter and returns calibrated concern probabilities per regulatory category. These are AI estimates, not regulatory advice.

• Q1. How does the OCA-Ped software's machine learning model differentiate between various stages or severities of carious lesions in pediatric dentition, and does its performance vary accordingly?; Given the slight decrease in specificity (1.1%) with OCA-Ped assistance in the MRMC study, what are the potential clinical implications of false positives, particularly in a pediatric population where conservative treatment is often prioritized?; What were the specific criteria used by the three general dentists to establish the consensus reference standard for caries detection in both the MRMC and standalone performance studies?; How does the device's performance (sensitivity, specificity, Dice coefficient) vary across different types of primary versus permanent teeth within the 4-11 age range, and are there more granular performance data across different radiographic sensor brands beyond the listed sensitivities?; Were there any specific types or locations of caries (e.g., interproximal, occlusal, early enamel lesions vs. more advanced dentinal caries) where the OCA-Ped demonstrated notably higher or lower accuracy compared to unaided visual inspection by dentists?