AI model detects osteoporosis in routine CT scans

A new AI model can detect low bone density and signs of osteoporosis in CT scans taken for other medical reasons, potentially improving early diagnosis and fracture prevention. The research was published in Radiology. The deep learning tool was applied to over 538,000 CT exams from 283,000 patients, with radiologists confirming more than 99% of its findings. This opportunistic screening approach aims to address the fact that only one in five eligible patients currently undergo osteoporosis testing. Osteoporosis causes over two million fractures annually in the U.S., leading to significant illness and death. Researchers established diagnostic thresholds for each vertebral body using WHO prevalence data, emphasizing that preventing fractures is far better than treating their consequences.