SYSTEM AND METHOD TO PREDICT CRANIOFACIAL AND DENTAL ABNORMALITIES USING CONVOLUTIONAL NEURAL NETWORKS

Abstract

This paper presents an advanced system for dental insurance treatment authorization that leverages AI-based tools to streamline the processing and analysis of dental x-rays and composite images. The system includes multiple specialized AI engines designed for specific tasks: the Landmark Localizer for identifying dental landmarks, the Ectopic Eruption Discoverer for detecting anomalies, and the Smart Decomposer, Composer, and Anonymizer modules for handling composite images. Operating in either single-case or batch mode, the system enables efficient, accurate analysis and reporting across multiple patient cases. Each AI engine functions autonomously yet integrates seamlessly within the system, automatically recognizing asset types and directing them to appropriate processing engines. Key innovations include auto-orientation of x-rays, anomaly marking, image decomposition, and re-composition capabilities that cater to diverse dental insurance verification needs. This approach reduces processing time, enhances

Specification

Description:[01] FIELD OF THE INVENTION
The present invention relates to the system and method to predict craniofacial and dental abnormalities using convolutional neural networks.
[02] BACKGROUND OF THE INVENTION
Since the mid-1990s, an extensive collection of dento-craniofacial visual assets (DCVA) has been steadily accumulating across a range of repositories, including those managed by universities, government bodies, insurance firms, research organizations, publishers, and even small clinics. Major internet search engines, such as Google®, have facilitated broader access to this type of visual data through online searches.
However, a significant challenge for the custodians and users of these DCVA collections is the absence of an automated system that can process multiple repositories, recognize content, generate descriptive labels and metadata, and optionally save this metadata in searchable formats. Such a system is essential to enable more intelligent and relevant responses to diverse user inquiries. Another majo , Claims:1. Input Module:
• A system that accepts input images in two modes:
• Single case-by-case mode for individual patient assessments.
• Batch mode for processing multiple patient images simultaneously.
2. Asset Recognition Engine:
• An engine that automatically identifies the type, category, and class of dental images (such as x-rays and composites) and routes them to the appropriate artificial intelligence (AI) processing engines for analysis.
3. Landmark Localizer: A dedicated AI engine that:
• Accepts single or batch x-ray inputs.
• Automatically determines the orientation of x-rays and flips left-side images to a standard view.
• Recognizes and localizes dental landmarks crucial for treatment authorization.
• Overlays the localized landmarks on the original x-ray images for visualization.
4. Ectopic Eruption Discoverer: A system component that:
• Processes x-ray images to identify ectopic eruptions and other dental anomalies.
• Marks these specific conditions on a copy of the original x-ray while leaving nor

Documents