Hip fractures are debilitating and are expected to increase in the United States over the coming years. Accurate and timely diagnoses for hip fractures reduce the time patients wait for operative interventions, a critical step in lowering morbidity and post-operative infections.
Across all departments at the University of California, San Francisco (UCSF), the Department of Radiology and Biomedical Imaging and the Center for Intelligent Imaging (ci2) analyze approximately 40-50 pelvis XR exams daily. Many hip fractures are first diagnosed at this stage.
The Hip Fracture Detection Program, managed by UCSF ci2's Felix Liu, MS, uses models aided by machine learning to improve physician accuracy in detection and classification. The clinical deployment of the AI-model is currently limited to a select group of Musculoskeletal radiology fellows for research purposes, but for older adults or anyone suffering from bone density loss caused by osteoporosis, this program will be instrumental in early detection and prevention.
Researchers, including Felix Liu, MS, John Mongan, MD, PhD, Christopher Hess, MD, PhD, James Hawkins, Rina Patel, MD, Zehra Akkaya, Beck Olson, Jason Crane, PhD, Thomas Link, MD, PhD, and Sharmila Majumdar, PhD, studied if deep learning methods could intervene in hip fracture. The authors present their findings in "Automatic Hip Fracture Identification and Functional Subclassification with Deep Learning," published in RSNA Radiology: Artificial Intelligence.
Visit the UCSF ci2 blog to read more about research and news.
