Accuracy of developing AI program to diagnose lumbar spinal canal stenosis with abdominal CT is 84%
Dec 02, 2024
Professor Lee Joon-woo and Lee Young-joon of the Department of Radiology at Bundang Seoul National University Hospital developed an AI program to diagnose lumbar spinal stenosis using computed tomography (CT).
The diagnostic accuracy of this algorithm is 84%, which is the level of a radiologist, which is fully available in clinical practice and is expected to help patients with limited use of magnetic resonance imaging (MRI).
Lumbar spinal canal stenosis is a disease that compresses the spinal cord or neuromuscular muscles as the spinal canal in the lumbar part narrows. It occurs in about 30% of people over the age of 60 in Korea, and the incidence rate is higher with the elderly. The main symptoms are back pain and numbness in legs and hips, and in severe cases, the nerve function that controls the lower body is impaired, which can lead to urination and bowel movements.
The problem is that symptoms of lumbar spinal canal stenosis are similar to 'intervertebral disc escapism (back disk)', a disease that is usually improved with conservative treatment, making it difficult to accurately diagnose with symptoms alone. In addition, MRI should be mainly used for diagnosis, but it was limited to patients who transplanted metal such as 'spinal nerve stimulator' or 'heart rate controller', so lumbar CT had to be taken.
In response, Professor Lee Joon-woo and Professor Lee Young-joon conducted a study to develop a program to diagnose lumbar spinal stenosis using abdominal CT. In addition to spinal diseases, abdominal CT is a widely used test method for clinical examination, and has the advantage of being cheaper and not affected by metal than MRI.
The research team developed an AI program using information on 109 patients who took abdominal CT and lumbar CT.
This program analyzed CT of patients and automatically classified them as 'stenosis' if the intra-lumbar dural sac was less than 100 ㎟. As a result, the diagnostic accuracy of the algorithm using abdominal CT was 84%, which was similar to the level of diagnosing lumbar CT by a radiologist. In particular, the diagnostic accuracy of severe spinal canal stenosis was very high, over 85%, and asymptomatic and mild lumbar spinal canal stenosis, which is difficult to diagnose with CT, could also be diagnosed.
If the program developed by this study is applied to clinical practice, it is expected that abdominal CT alone can diagnose lumbar spinal stenosis. In addition, abdominal CT is also used in other disease tests such as health checkups, so if it is used, it is expected that lumbar spinal canal stenosis can be checked without a separate test.
Professor Lee Jun-woo "Abdominal CT was developed with the idea that it is the most common imaging test used to examine the abdomen and internal organs, and that the spinal area is also photographed.""If you use this, you can simply diagnose lumbar spinal stenosis with an abdominal CT alone."," he said.
Professor Lee Young-joon of the Department of Radiology then said, "This study proved that the AI program can diagnose lumbar spinal stenosis with accuracy close to that of a radiologist."We will strive to develop an integrated diagnostic program that encompasses all diseases occurring in the spine in addition to lumbar spinal stenosis."," he said.
Meanwhile, the study was published in the official collection of papers published by the International Musculoskeletal Society 'Sculoskeletal Radiology'.
The diagnostic accuracy of this algorithm is 84%, which is the level of a radiologist, which is fully available in clinical practice and is expected to help patients with limited use of magnetic resonance imaging (MRI).
Lumbar spinal canal stenosis is a disease that compresses the spinal cord or neuromuscular muscles as the spinal canal in the lumbar part narrows. It occurs in about 30% of people over the age of 60 in Korea, and the incidence rate is higher with the elderly. The main symptoms are back pain and numbness in legs and hips, and in severe cases, the nerve function that controls the lower body is impaired, which can lead to urination and bowel movements.
The problem is that symptoms of lumbar spinal canal stenosis are similar to 'intervertebral disc escapism (back disk)', a disease that is usually improved with conservative treatment, making it difficult to accurately diagnose with symptoms alone. In addition, MRI should be mainly used for diagnosis, but it was limited to patients who transplanted metal such as 'spinal nerve stimulator' or 'heart rate controller', so lumbar CT had to be taken.
In response, Professor Lee Joon-woo and Professor Lee Young-joon conducted a study to develop a program to diagnose lumbar spinal stenosis using abdominal CT. In addition to spinal diseases, abdominal CT is a widely used test method for clinical examination, and has the advantage of being cheaper and not affected by metal than MRI.
The research team developed an AI program using information on 109 patients who took abdominal CT and lumbar CT.
This program analyzed CT of patients and automatically classified them as 'stenosis' if the intra-lumbar dural sac was less than 100 ㎟. As a result, the diagnostic accuracy of the algorithm using abdominal CT was 84%, which was similar to the level of diagnosing lumbar CT by a radiologist. In particular, the diagnostic accuracy of severe spinal canal stenosis was very high, over 85%, and asymptomatic and mild lumbar spinal canal stenosis, which is difficult to diagnose with CT, could also be diagnosed.
If the program developed by this study is applied to clinical practice, it is expected that abdominal CT alone can diagnose lumbar spinal stenosis. In addition, abdominal CT is also used in other disease tests such as health checkups, so if it is used, it is expected that lumbar spinal canal stenosis can be checked without a separate test.
Professor Lee Jun-woo "Abdominal CT was developed with the idea that it is the most common imaging test used to examine the abdomen and internal organs, and that the spinal area is also photographed.""If you use this, you can simply diagnose lumbar spinal stenosis with an abdominal CT alone."," he said.
Professor Lee Young-joon of the Department of Radiology then said, "This study proved that the AI program can diagnose lumbar spinal stenosis with accuracy close to that of a radiologist."We will strive to develop an integrated diagnostic program that encompasses all diseases occurring in the spine in addition to lumbar spinal stenosis."," he said.
Meanwhile, the study was published in the official collection of papers published by the International Musculoskeletal Society 'Sculoskeletal Radiology'.
|
bellho@sportschosun.com
