Study: AI improves stroke, dementia diagnosis in brain scans

Study data published in the journal Radiology suggest that machine learning can identify and measure on computed tomographic (CT) images the severity of small vessel disease (SVD), one of the commonest causes of stroke and dementia. Paul Bentley, lead author, remarked that "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." He added that "our technique is consistent and achieves high accuracy relative to an MRI scan…This could lead to better treatments and care for patients in everyday practice."

The retrospective study looked at 1082 CT scans of patients with acute ischaemic stroke from 2000 to 2014, including a subset from the third international stroke trial (IST-3). The researchers compared their automated method for segmenting cerebral white matter lesions (WMLs), a small vessel disease marker, on CT images to fluid-attenuated recovery magnetic resonance (MR) images and expert consensus.

The software identified and measured WML volumes, providing a score ranging from mild to severe disease.  Results were then compared to a panel of doctors who estimated small vessel disease severity from the same CT scans. According to study findings, the level of agreement of the software with the doctor was as good as agreements between one doctor with another.

In addition, both MR images and CT scans were obtained from 60 subjects. The researchers explained that when the MRI scans were used to estimate the exact amount of small vessel disease, the software was found to be 85-percent accurate at predicting the severity of small vessel disease. The scientists are now using similar methods to measure the amount of brain shrinkage and other conditions commonly diagnosed on brain CT.

Joanna Wardlaw, study author, noted that "this is a first step in making a scan reading tool that could be useful in mining large routine scan datasets and, after more testing, might aid patient assessment at hospital admission with stroke." Meanwhile, Bentley remarked "the importance of our new method is that it allows for precise and automated measurement of the disease. This also has applications for widespread diagnosis and monitoring of dementia, as well as for emergency decision-making in stroke."

Did you like this article?