Machine learning shown to detect one of the commonest causes of dementia and stroke more accurately than current methods
New software developed by scientists at Imperial College London and Edinburgh University has been able to identify and measure the severity of small blood vessel disease, one of the commonest causes of stroke and dementia.
The researchers say the technology could help clinicians to administer the best treatments more quickly in emergency settings, and also be used to predict a person’s likelihood of developing dementia.
“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 computed tomography (CT) scanning,” said Paul Bentley, lead author and clinical lecturer at Imperial College. The technique is consistent and achieves high accuracy relative to magnetic resonance imaging (MRI), the current gold standard technique for diagnosis, according to Bentley.
Small vessel disease (SVD), a very common neurological disease in older people, reduces blood flow to the deep white matter connections of the brain, damaging and eventually killing the brain cells. It causes stroke and dementia as well as mood disturbance. SVD increases with age but is accelerated by hypertension and diabetes.
At the moment, SVD is diagnosed by looking for changes to white matter in the brain during MRI or CT scans. However, this relies on gauging from the scan how far the disease has spread. In CT scans it is often difficult to decide where the edges of the SVD are, making it difficult to estimate the severity of the disease.
Although MRI can detect and measure SVD more sensitively, it is not the most common method used due to scanner availability and suitability for emergency use or for older patients.
“Current methods to diagnose the disease through CT or MRI scans can be effective, but it can be difficult for doctors to diagnose the severity of the disease by the human eye,” Bentley said. “The importance of our new method is that it allows for precise and automated measurement of the disease.”
This has relevance for the diagnosis and monitoring of dementia and for emergency decision-making in treating stroke patients.
In stroke, clot-busting drugs must be administered within three hours of onset of the seizure. However, these treatments also pose the risk of causing brain haemorrhages, and the likelihood this will occur becomes more likely as the amount of SVD increases. The software could be applied to estimate the likely risk of haemorrhage.
In addition, the software could help assess the likelihood of patients developing dementia or immobility due to slowly progressive SVD. Doctors could also be altered to potentially reversible causes of SVD, such as high blood pressure or diabetes.
The study, published this month in Radiology, used historical data of 1,082 CT scans of stroke patients across 70 hospitals in the UK between 2000-2014. The software identified and measured a marker of SVD and gave a disease severity score.
The researchers then compared the results to those of a panel of expert doctors who estimated SVD severity from the same scans. The level of agreement of the software with the experts was as good as the level of agreement between one expert and another.
In 60 cases the researchers obtained MRI and CT scans of the same subjects and used the MRI image to estimate the exact amount of SVD. This showed that the software is 85 per cent accurate at predicting the severity of SVD.