News Release

Neural network-assisted bone fracture detection

Peer-Reviewed Publication

Proceedings of the National Academy of Sciences

A Heatmap Overlaid on the Radiograph

image: A heatmap overlaid on the radiograph. When the model determines that a fracture is present, the heatmap represents the model's confidence that a particular location is part of the fracture, with yellow and blue being more and less confident, respectively. view more 

Credit: Image courtesy of Imagen Technologies.

Researchers report that a deep neural network designed to identify fractures in radiographs that was trained using 135,409 annotated radiographs from 18 senior orthopedic surgeons helped emergency department clinicians reduce the rate of radiograph misinterpretation by 47%, a result demonstrating how deep neural networks, trained by experienced specialists, can transfer expertise to generalist clinicians.

###

Article #18-06905: "Deep neural network improves fracture detection by clinicians," by Robert Lindsey et al.

MEDIA CONTACT: Robert Lindsey, Imagen Technologies, New York, NY; tel: 580-279-9987; e-mail: rob@imagen.ai


Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of news releases posted to EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.