Would you trust an artificially intelligent doctor?
AI is reading CT scans to more efficiently detect bone fractures in osteoporosis patients
CT scans enable doctors to see 3-D X-ray images of inside the body. Now, artificial intelligence (AI) algorithms are successfully reading CT scans to spot bones at risk of breaking – even if the scan was originally taken to look for something completely different.
Spine fractures are a symptom of osteoporosis, a condition that makes bones more fragile and likely to break. Often, patients with this disease can suffer fractures without even noticing. However, it’s important to spot them and begin treatment as early as possible, before they lead to bigger issues and more broken bones.
Nurses in Oxford, United Kingdom recently used the Zebra Imaging Analytics Engine to re-analyze the CT scans of nearly 5000 patients. They were all classified as either positive (signs of fracture) or negative (no fracture). The nurses verified each result and found that the AI didn’t miss a single fracture. It did err on the side of caution though – about 30% of the scans that were reported as positive by AI were actually nothing to worry about. A similar study has also recently been done by researchers at Dartmouth College.
Overall, the software saves a lot of time for the nurses, as they can now just review the positive scans instead of studying every scan in detail. It has helped identify over 100 patients with previously undetected fractures. Thanks to AI, these people have been able to receive early treatment for osteoporosis, protecting them against broken bones caused by the disease.