Dental Arch Image Analysis for Remote Orthodontic Assessment

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current orthodontic treatments require patients to send images to orthodontists for assessment, which is inconvenient and inefficient.

Innovation Solution

A method using deep learning devices, preferably neural networks, to analyze dental arch images by creating a training set of historical images, training the network, and determining tooth and image attributes, with a process to enrich the training database by acquiring updated images under varying conditions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If a deep learning device is trained using a large training set of historical images, then the analysis precision and reliability are improved, but the time and resources required for training increase

Engineering Contradiction:
Improveanalysis precisionVSAvoidtraining time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent applies preliminary action by pre-training the deep learning device using a large training set of historical images before actual use. The training set comprises multiple images of dental arches with annotated tooth attributes, and the deep learning device is trained in advance to recognize and analyze these attributes. This allows the system to achieve high analysis precision without requiring extensive training time during actual deployment, as the training work has been performed beforehand.

Inventive Principle:
Principle #10Preliminary action

2Adaptability or versatility

If the training set is enriched with diverse images under varying acquisition conditions, then the adaptability and robustness of the analysis are improved, but the complexity of data collection and processing increases

Engineering Contradiction:
ImproveadaptabilityVSAvoiddata collection complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent applies universality by creating a training set that includes images captured under various acquisition conditions (different devices, angles, lighting, and environments). The deep learning device is trained to handle this diversity, making it universally applicable to different real-world scenarios. This enables the system to adapt to various imaging conditions without requiring separate models for each condition, thereby improving adaptability while managing complexity through a unified training approach.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Solution Approach 2:

The patent applies parameter changes by varying multiple parameters during image acquisition for the training set, including acquisition device type, viewing angle, lighting conditions, and image quality. By training the deep learning device on images with these varied parameters, the system learns to recognize tooth attributes across different conditions. This approach improves adaptability by teaching the model to handle parameter variations rather than requiring separate systems for each condition.

Inventive Principle:
Principle #35Parameter changes

3Productivity

If automated analysis is implemented using deep learning, then productivity and efficiency are improved, but the device complexity and computational resources required increase

Engineering Contradiction:
Improveanalysis efficiencyVSAvoidsystem complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent applies self-service by implementing automated analysis where the deep learning device independently processes images of dental arches without requiring manual intervention. The system automatically detects tooth areas, extracts attributes (such as tooth type, position, and condition), and generates analysis results. This automation significantly improves productivity and efficiency compared to manual analysis by orthodontists, while the complexity is managed through the use of established deep learning frameworks and algorithms.

Inventive Principle:
Principle #25Self-service

4Loss of information

If the deep learning device determines multiple tooth attributes simultaneously, then the comprehensiveness of the analysis is improved, but the computational complexity and processing time increase

Engineering Contradiction:
Improveanalysis comprehensivenessVSAvoidcomputational complexity
Core Design Contradiction:
Loss of informationVSDevice complexity

Solution Approach 1:

The patent applies segmentation by dividing the analysis process into distinct components: first identifying tooth areas in the image, then determining multiple attributes for each detected tooth (such as tooth type, position, condition, and other characteristics). This segmented approach allows the deep learning device to handle multiple attributes systematically, improving comprehensiveness while managing computational complexity by processing attributes in an organized sequence rather than attempting simultaneous analysis of all parameters.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP4647036A1Method for analysing an image of a dental arch
Publication Date: 2025.11.12 DENTAL MONITORING
  • EP4647036A1 patent drawingFigure 1~2
  • EP4647036A1 patent drawingFigure 3~4
  • EP4647036A1 patent drawingFigure 5~6

AI summary

A method for analyzing an image of a patient's dental arch, referred to as the "analysis image," comprising the acquisition of the analysis image by the patient using a mobile phone, the analysis image being in color and chosen from a photograph and an image extracted from a film, the mobile phone being held more than 5 cm away from the dental arch during acquisition, said analysis method comprising the following steps: 1') creation of a training database containing more than 1,000 images of dental arches, or "historical images," each historical image containing an attribute value for an image attribute, or "image attribute value"; 2') training at least one neural network, using the training database, by presenting the historical images as input to the neural network, so that the neural network gradually learns to recognize patterns in an image and to associate them with image attribute values;3') submitting the analysis image to the neural network so that it determines, for said analysis image, at least one probability relative to said image attribute value, and determining, as a function of said probability, a value for said image attribute for the analysis image; said image attribute being relative to - a position and/or an orientation and/or a calibration of the acquisition device used to acquire said analysis image, or - a quality of the analysis image, or - the brightness, contrast or sharpness of the analysis image,;