Assessment of Bone Age using Artificial Intelligence in children and adolescents across different ethnicities

Purpose and goal:

The project aimed to automate the assessment of skeletal age in children and adolescents. The goal was to develop a user-friendly AI-based software for fast, cost-effective and safe assessment of skeletal age in children and adolescents of different ethnicities. The project, which was run in collaboration with an Indian partner, succeeded in achieving its aim.

As a continuation of the project, the software will be further developed to include a mobile application to increase user-friendliness and accessibility. Expected results and effects:

We have developed and trained an AI-based algorithm for assessment of skeletal age and compared the result with other existing solutions. Currently, our solution is broadly comparable to non-AI-based manual assessment of skeletal age performed by trained people. Being able to make the assessment of skeletal age automated with the help of AI may increase the availability and lower the costs of this important analysis, which is currently not available in large parts of the world.

Approach and implementation:

The project´s structure consisted of a collaboration between Sweden and India, where Karolinska Institutet has extensive knowledge of the use of AI and assessment of skeletal age images. The Indian team has great expertise in the development of AI-supported mobile applications and has also had the opportunity to validate developed software on a very extensive image material that was successfully analyzed. The project had a somewhat slow start due to the pandemic, but has since run according to plan.