Deep vein thrombosis (DVT) is a kind of blood clot that causes swelling, pain, and discomfort. Nearly 8 million people worldwide potentially have a venous blood clot every year. And, 30-50% of those who develop a DVT can go on to have long-term symptoms and disability, according to a study by Oxford University.
(Sign up to our Technology newsletter, Today's Cache, for insights on emerging themes at the intersection of technology, business and policy. Click here to subscribe for free.)
So, a team of researchers at Oxford University, Imperial College and the University of Sheffield, have designed an algorithm to diagnose DVT like a radiologist. Usually, a trained radiographer performs an ultrasound scan to diagnose DVT.
“We have found that the preliminary data using the AI algorithm, coupled with a hand-held ultrasound machine shows promising results,” Dr Nicola Curry, the study’s lead and a clinician at Oxford University Hospitals NHS Foundation Trust, said in a statement.
The team collaborated with a tech firm to train an algorithm to spot patients with DVT. In preliminary testing, it found the algorithm to accurately diagnose DVT when compared to the ultrasound scan, the university noted.
The algorithm can also help people find the right “locations along the femoral vein, so that even a non-specialist user can acquire the right images,” Christopher Deane, a study team member, said.
This can enable non-specialist healthcare professionals, like general physicians and nurses, to diagnose and treat DVT faster. Currently, due to the absence of a definitive diagnosis within 24 hours of a suspected DVT many patients end up receiving painful injections, Dr Curry explained.
The research team plans to start a test-accuracy blinded clinical study, that will check the AI’s accuracy and assess its ability to identify DVT cases. They have detailed their work in an article titled ‘Non-invasive diagnosis of deep vein thrombosis from ultrasound imaging with machine learning,’ published in the journal Digital Medicine .