|AI Brain scan. Credits: Imperial College London|
A prominent economist known for her stern views and defiant stance against the government’s war on drugs was being interviewed on TV. In between her profuse tirades, her right cheek drooped, her words slurred and weak. She was having a stroke. Had she not been cut off on air and rushed to ER, she could have not lived to tell the tale.
A stroke happens when the blood vessels serving oxygen to a part of the brain is cut off, blocked or burst open. The brain cells get fewer oxygen and start to die. As a result, the part of the body controlled by the dying brain cells would stop working.
|Source: Pexels Creative Commons|
The World of Strokes
This life-threatening condition ranks 2nd leading cause of death of 6.2 million people worldwide in 2016. Those who survived the stroke were, at some degree, disabled.
The World Health Organization (WHO) reported that annually, there are more deaths caused by stroke than deaths related to AIDS, Tuberculosis, and Malaria combined.
According to Center for Disease Control and Prevention (CDC), stroke affected 795,000 people in the US every year. About 130,000 of them die. Translated into every setting, stroke happens to someone every 40 seconds and kills someone every 3 minutes and 45 seconds. It is the leading cause of long-term disability in the US.
These figures could be reduced with the use of AI to accurately and quickly diagnose the severity of brain attack at the critical time. Stroke is preventable and treatable.
Scientists at the Imperial College London and the University of Edinburg in Britain recently unveiled a machine-learning software created to identify and measure the state of small vessel disease (SVD) in the brain. This Artificial Intelligence is faster, more accurate than the current gold standard technique in the diagnosis of stroke.
SVD is the most common cause of stroke and dementia. This neurological disease is common in older people between 60-90 years old. SVD refers to a variety of abnormalities in the small vessels in the brain. Brain cells actually appear white in MRIs. When blood supply is reduced, changes in the deep, white connections in the brain happen. SVD can be small strokes (lacunar infarcts), can be bleeding of the brain from very small blood vessels (cerebral microbleeds), or white matter hyperintensities.
Dr. Paul Bentley, a clinical lecturer at Imperial College London and lead author of the research said, "This is the first time that machine learning methods have been able to accurately measure a marker of small vessel disease in patients suspected with stroke or memory impairment and who undergo CT scanning."
The current methods of diagnosis involved MRI or CT scan which showed changes to the white matter in the brain. Gauging how far SVD has affected the brain is quite difficult in CT Scans. Doctors cannot tell where the edges of the SVD are, therefore, they cannot accurately determine the severity.
MRIs, on the other hand, are much more sensitive and reliable. These can detect and measure SVD but the availability during emergency setting is an issue. Also, MRIs may not be suitable for older patients.
Bentley added, "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. The importance of our new method is that it allows for precise and automated measurement of the disease."
|Source: Pixabay Creative Commons|
The AI Advantage
When a patient is rushed to the ER due to stroke, clot-busting drugs must be administered within 3 hours from the onset of the seizure. However, these treatments could cause brain hemorrhages. Without AI, the risks of hemorrhages cannot be determined. Thus, brain hemorrhages could occur while severity of SVD increases.
The research team also explained that the software can statistically determine the likelihood of patients developing dementia, or immobility due to slow, progressive SVD. With AI, doctors could shift to medications to treat causes of SVD such as high blood pressure or diabetes. In emergency settings, the AI could aid the attending physicians with quick diagnosis and in administering the best possible treatments.
The test results conducted at Charing Cross Hospital (part of Imperial College Healthcare National Health Service Trust) were encouraging. The data involved 1,082 CT scans of stroke patients in 70 hospitals, including cases from the Third International Stroke Trial between 2000 and 2014.
Based on the data, the learning machine software identified and measured an SVD marker. It was able to score the severity of the disease ranging from mild to severe. The AI results were compared to the information from a panel of expert doctors who gauged the severity of SVD based on the same scans. Their estimates and the AI data were comparable. The AI results appeared as though they came from the experts.
The AI was also accurate by 85% at predicting the severity of SVD from 60 MRI and CT scans of the same subjects.
Dr. Joanna Wardlaw, head of neuroimaging sciences at the University of Edinburgh emphasized 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."
The research study was recently published in the journal Radiology.