Home » News Detail

News Detail

Use of Artificial Intelligence Program to Diagnose DVT

  • 23
    By Katherine Kozlowski, Medical Author and Contributor to Vein News and VeinsVeinsVeins.com


    Researchers from the United Kingdom and Europe trained an Artificial Intelligence (AI) program that is designed to diagnose DVT with the use of an app and an ultrasound machine.


    The current gold standard of diagnosis is compression ultrasound done by specially-trained sonographers.


    The researchers used ultrasound videos from 255 patients to train the AutoDVT program. They then evaluated the performance of the program by measuring the accuracy on 83 patients.


    They found that the AutoDVT program was able to correctly diagnose DVT 82-94% of the time, and able to correctly identify a negative result 70-82% of the time.


    They also completed a cost-benefit analysis compared to the current gold standard of care and found that using the AI program has the potential to save 71-139 GBpounds ($97-$191 USD) per exam.


    The development of this software could more quickly and efficiently diagnose DVT in the future. Further clinical research is planned for the future to evaluate the product further.




    Mainz B, Heinrich M, Makropoulos A et al. Non-invasive diagnosis of deep vein thrombosis from ultrasound imaging with machine learning. npj Digital Medicine 2021;4(1): 1-18.