A variety of readers were used to simulate real-life scenarios, including radiologists, orthopedic surgeons, emergency physicians and physician assistants, rheumatologists, and family physicians, all of whom read x-rays in real clinical practice to diagnose fractures in their patients. Each reader's diagnostic accuracy of fractures, with and without AI assistance, were compared against the gold standard. They also assessed the diagnostic performance of AI alone against the gold standard. AI assistance helped reduce missed fractures by 29% and increased readers' sensitivity by 16%, and by 30% for exams with more than 1 fracture, while improving specificity by 5%.

Emergency room and urgent care clinics are typically busy and patients often have to wait many hours before they can be seen, evaluated and receive treatment. Waiting for x-rays to be interpreted by radiologists can contribute to this long wait time because radiologists often read x-rays for a large number of patients. A new study has found that artificial intelligence (AI) can help physicians in interpreting x-rays after an injury and suspected fracture. "Our AI algorithm can quickly and automatically detect x-rays that are positive for fractures and flag those studies in the system so that radiologists can prioritize reading x-rays with positive fractures. The system also highlights regions of interest with bounding boxes around areas where fractures are suspected. This can potentially contribute to less waiting time at the time of hospital or clinic visit before patients can get a positive diagnosis of fracture," explained corresponding Ali Guermazi, MD, PhD, chief of radiology at VA Boston Healthcare System and Professor of Radiology & Medicine at BUSM.

Machine learning using deep convolutional neural networks (CNNs) can be used to detect fractures in plain radiographs, according to a new study published in Clinical Radiology. A team of researchers from the U.K. taught the CNNs using lateral wrist radiographs performed at a single facility from January 2015 to January 2016. Each image was classified as "fracture" or "no fracture" based on the existing radiology report. The distinction was personally verified by a human specialist before data was used to "train" the CNN. Overall, the area under the receiver operator characteristic curve (AUC) was 0.954, a number the authors said provided a proof of concept.