Researchers at Imperial College London and the University of Edinburgh in Scotland developed a machine learning software able to detect small vessel disease, a common cause of dementia and stroke, in brain CT scans with almost 90 percent accuracy.
The researchers' findings, published in Radiology, may help physicians better predict a person's risk of developing dementia and administer treatment more quickly during an emergency.
“This is the first time that machine learning methods have been able to accurately measure a marker of small vessel disease in patients presenting with stroke or memory impairment who undergo CT scanning," said lead author Paul Bentley, PhD, a clinical lecturer at Imperial College London, in a prepared statement. "Our technique is consistent and achieves high accuracy relative to an MRI scan—the current gold standard technique for diagnosis. This could lead to better treatments and care for patients in everyday practice.”
The software can pinpoint and diagnose small vessel disease (SVD), a common neurological disease that reduces blood flow to white matter connections in the brain, kills brain cells and causes stroke or dementia. Although the disease can be diagnosed through CT or MRI, the severity of SVD isn't as easy to measure with these methods.
"The importance of our new method is that it allows for precise and automated measurement of the disease [SVD]," Bentley said. "This also has applications for widespread diagnosis and monitoring of dementia, as well as for emergency decision-making in stroke.”
Bentley and his colleagues collected the data of 1,082 CT scans of stroke patients from 70 different hospitals across the U.K. between the years 2000 to 2014.
After the machine learning software identified and measured SVD markers and determined its severity, the researchers compared their results to those from panel of clinical experts who estimated SVD severity from CT scans alone.
The software was 85 percent accurate at predicting the severity of SVD, the researchers wrote.