Training machines to diagnose human injuries
DRIVEN BY DATA
Through its Data + Decisions Destination Area, Virginia Tech inventively interweaves data science into its curriculum. Meet a faculty member using data in novel and world-changing ways.
Musculoskeletal injuries, such as small tendon tears, can be challenging for the human eye to detect on ultrasound images.
Vincent Wang, the Kevin P. Granata faculty fellow and associate professor in biomedical engineering and mechanics, is using clinical ultrasound images to train computers to detect these injuries, with the goal of facilitating more accurate medical diagnoses.
Carrie Cheung, a graduate student in biomedical engineering, works with Wang to develop algorithms to identify ultrasound image features unique to injured tendons. Their hope is that these algorithms can be used in clinical settings where machines can identify injuries in real-time. These analyses may assist with clinical diagnosis and injury prevention.
“Our approach resembles that used for facial recognition in commonly used smartphone apps,” Wang said.
This project is a collaboration with Bert Huang and Wu Feng in computer science at Virginia Tech for creation of algorithms and code and Albert Kozar in sports medicine at the Edward via College of Osteopathic Medicine to supply the tendon images.
Watch the video to learn more about the research.
- Written by Laura Weatherford