Funding outcome

One of the most common reasons a child visits the emergency room (ER) is arm injuries, especially wrist and elbow fractures. Active children with developing skeletons are particularly prone to these injuries. Traditional diagnosis via radiographs requires a long visit to the ER, often waiting many hours to have x-rays performed with specialized equipment. Ultrasound, using portable and low-cost devices, has been shown to be successful in children's wrist and elbow fracture detection. However, ultrasound images can be tricky to interpret, and require expert analysis. Therefore, we aim to develop an artificial intelligence (AI) tool for the automatic diagnosis of children's wrist and elbow fractures. So far we have collected ultrasound videos from 211 children with wrist injuries and 116 children with elbow injuries aged 0-17 from Stollery Children's Hospital in an ethics-approved study. Using these images, we plan to build an AI tool with three stages: 1. Find images showing the body parts we need to evaluate, 2. Automatically identify the bones in each of these images, and 3. Identify fractures of these bones. The AI will be integrated into a software app on Android phones/tablets, and validated in pediatric ER in Alberta. By connecting a pocket-size ultrasound probe to a cell phone installed with our App, lightly trained users can use this portable tool to assess arm injuries, which will be a useful tool in rural and remote areas. It would be particularly helpful as only patients with suspected fractures need to travel to the hospital ER for follow-up examination and treatment, relieving the burden on the ER and providing equitable access to children's diagnostic care.