18 hours Ago By Korea Bizwire
Researchers have created an artificial intelligence model that can diagnose lumbar spinal stenosis using standard X-ray images, potentially replacing the need for expensive and time-consuming MRI scans. (Image created by AI/ChatGPT) SEOUL, April 3 (Korea Bizwire) — In a breakthrough that could broaden access to spinal care, South Korean researchers have developed an artificial intelligence model capable of diagnosing lumbar spinal stenosis using standard X-ray images—eliminating the need for costly and time-intensive MRI scans. A research team led by Professor Chang-Hyun Lee from the Department of Neurosurgery at Seoul National University Hospital announced on Wednesday that it had trained the AI model using X-ray images of 2,500 patients diagnosed with lumbar spinal stenosis and 2,500 individuals without the condition, all taken between 2005 and 2017.
Lumbar spinal stenosis, a condition in which the spinal canal narrows and compresses nerves, commonly causes pain, numbness, or weakness in the lower body. While MRI remains the gold standard for diagnosis, its high cost, long imaging time, and limited availability—particularly in smaller clinics—have posed accessibility challenges for patients. To address this, the team fed the AI model X-ray images taken in three distinct postures: neutral, flexion (bending forward), and extension (leaning backward).
By analyzing and integrating data from these varied positions, the model improved diagnostic accuracy while reducing false positives and missed detections. South Korean researchers have developed an artificial intelligence model capable of diagnosing lumbar spinal stenosis using standard X-ray images. (Image courtesy of Seoul National University Hospital) The AI demonstrated a diagnostic accuracy of 91.
4% in internal tests and 79.5% in external validation. Researchers anticipate that the tool could significantly enhance early detection in settings where MRI is impractical or unavailable.
X-ray imaging, being more affordable, faster, and portable, could also improve patient convenience. “With this model, spinal stenosis could be diagnosed through simple X-rays alone,” said Professor Lee. “For patients experiencing persistent but mild back pain, the AI tool could serve as a cost-effective screening method, potentially reducing the burden of unnecessary MRIs.
” The study was published in the international journal Scientific Reports and paves the way for further integration of AI into diagnostic radiology ahead of expected clinical applications in 2026. Ashley Song ( [email protected] ).
