Overjet Image Enhancement Assist

Overjet Image Enhancement Assist is an image processing software that employs a learning-based algorithm for noise reduction in bitewing/periapical radiographs and standard techniques for contrast/sharpness in all dental radiographs to improve image quality for viewing. This optional tool within the Overjet platform aims to enhance visualization for clinicians without altering the original images used for AI-driven diagnostic predictions.

Overjet Image Enhancement Assist is a Software as a Medical Device (SaMD) designed to improve the quality of dental radiographic images viewed within the Overjet device platform. It is intended for use by dental professionals as an optional tool to enhance bitewing, periapical, and panoramic images, supporting routine dental image review. The software does not replace the clinic's standard imaging acquisition or diagnostic systems but offers an improved visual display that can be toggled on or off by the user. Importantly, any artificial intelligence-based predictions for dental findings like caries or calculus are performed on the original, unenhanced images, ensuring that the image enhancement feature does not alter the primary diagnostic outputs of other Overjet devices. The device employs different techniques for image enhancement depending on the radiograph type. For bitewing and periapical images, it utilizes a learning-based algorithm specifically for noise reduction, complemented by standard, non-learning based techniques to enhance contrast and sharpness. For panoramic images, the software applies only these standard techniques for contrast and sharpness enhancement, without the learning-based noise reduction. Performance testing, including software verification, validation, and a retrospective study using quantitative metrics (like CNR and PSNR) and expert clinical evaluations, was conducted to demonstrate its image quality enhancement capabilities.

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 learning-based noise reduction algorithm for bitewing and periapical images compare in performance to traditional denoising techniques in terms of preserving fine diagnostic details?; What specific types or levels of noise is the learning-based algorithm most effective at reducing, and are there any known limitations or image characteristics where its performance might be suboptimal?; Can the degree of contrast and sharpness enhancement be customized by the user, or is it a fixed algorithm output for each image type?; Given that AI predictions for findings are run on unenhanced images, what is the primary clinical decision-making benefit anticipated from viewing the enhanced images, and how has this been validated?; What were the key metrics in the 'Likert expert clinical evaluation' used to assess image quality, and did the evaluation differentiate between the utility of noise reduction versus contrast/sharpness enhancement?