The orthopedics clinic is one of the busiest clinics at BC Children’s with over 14,000 patient visits each year. There are eight highly skilled orthopedic surgeons who provide care that changes the lives of kids with a variety of conditions and diseases including: traumatic injuries, spinal deformities, hip dysplasia, bone cancer, cerebral palsy, broken bones and athletic injuries. Pediatric elbow fractures are the most common injury seen in the orthopedics clinic and can range from severe injury that requires urgent surgery to minor injury that might only require a cast.

BC Children’s is the only hospital in the province devoted exclusively to the care of children – this means that kids and youth from full-grown 16-year-old teenagers to two-year-old toddlers all rely on BC Children’s for the specialized care they can’t receive anywhere else. As you can imagine, a child’s anatomy changes and evolves significantly as they grow which can sometimes lead to diagnostic inaccuracies when interpreting x-ray images, especially by those who do not specialize in pediatric radiology. These inaccuracies can potentially result in long-term complications. This is particularly prevalent in elbows injuries as there are several bones, growth plates, ligaments and tendons that could be affected.

There is a pressing need for an innovative diagnostic solution that is precise and user-friendly for healthcare providers across various specialties.

Artificial Intelligence (Al) offers a promising solution to this challenge. Machine learning can be likened to a highly intelligent assistant capable of identifying patterns within images. This Al-integrated approach is designed to make the diagnosis process more straightforward and precise, thereby empowering healthcare teams to make informed decisions with confidence. This project aims to harness this technology through a two-pronged approach to simplify and enhance the accuracy of diagnosing pediatric elbow fractures:

• • • Smart labeling: Utilizing Large Language Models (LLMs), this phase involves AI automatically recognizing and labeling critical features within elbow x-ray images. This process equips the system with the ability to highlight essential anatomical and pathological details, preparing the groundwork for in-depth analysis.
    • Fracture detection: The subsequent phase employs Convolutional Neural Networks (CNNs), renowned for their competence in image analysis, to accurately identify the presence and precise location of fractures. This capability provides clinicians with an AI-enhanced tool to detect fractures that might be easily overlooked, ensuring no detail is missed.

The integration of AI in diagnosing pediatric elbow fractures stands to revolutionize patient care. By offering a diagnostic tool that combines high accuracy with ease of use, the risk of misdiagnosis and the subsequent complications can be significantly reduced. Beyond improving care in pediatric orthopedics, this project also showcases the potential for wider application of AI in healthcare, leading to a future where advanced AI powered diagnostic capabilities are more accessible, even in areas with limited resources. This initiative is in line with emerging research that highlights the effectiveness of machine learning in medical imaging, marking a significant leap forward in the integration of AI technologies in pediatric healthcare