AI-powered service for creating ideal dental occlusion based on a client's photograph

AI-powered chatbots and the LLM model

solution

Technical solution

The service was deployed on a dedicated GPU server, ensuring high performance and reliability.
Full technical documentation (API description) and the service source code were provided.
The service was tested and integrated into the doctor's personal account.

Architecture:
Microservice architecture, separated into a backend service (receiving and processing requests)
and an AI processing service (executing the generation pipeline).

Programming Language:
Python

Frameworks and Libraries:
ComfyUI / LoRA, OpenCV computer vision libraries.

Infrastructure:
GPU server (RTX 3090) for hosting and executing the AI ​​processing service.

API:
REST API using the JSON format for data exchange.

Data Storage:
Temporary storage of original and final images on a local drive with automatic deletion after processing.

Automation:
Using Docker to containerize the application and simplify deployment and scaling.

Result

Business values

For orthodontists:

For patients:

Increased patient engagement:
Visualizing the expected treatment outcome motivates patients to begin and continue treatment.

Improved communication with patients:
The service allows for a more visual explanation of the treatment plan and expected results.

Increased treatment consent:
Visualizing a positive outcome increases patient trust and the likelihood of treatment consent.

Time savings:
Automating the visualization step saves physician time and allows them to focus on more complex tasks.

Realistic vision of the outcome:
Being able to see how your smile will change after treatment helps you make an informed decision.

Reduced anxiety:
Understanding the treatment process and expected results reduces anxiety and fear about orthodontic treatment.

Convenience and accessibility:
Visualizations are available remotely, without visiting the clinic.

solution

Results of implementation

The service is a REST API that accepts a user's photo as input and returns a processed image visualizing the orthodontic treatment results.

The processing pipeline includes the following steps:

Image acquisition:
The received image is assigned a unique ID, after which it is sent to the processing queue, and its status is changed to "in processing."

Face and smile detection:
Computer vision algorithms are used to detect faces and smiles in the image.

Bite correction and new smile generation:
Diffusion neural networks are used to correct bite and create an aesthetically pleasing smile.

Final image generation:
After processing, the status is changed to "ready," and the final image becomes available for retrieval.

Result regeneration:
A method is provided for regenerating the result.

Delete original files:
The original files are automatically deleted from the server after processing is complete and the result is presented to the user.

Development of software
and Big Data solutions

Send a request and our specialists will contact you within 1 hour.

Choose a convenient method of communication