Scientists have trained artificial intelligence to find COVID-19 on ultrasound images of the lungs

Scientists have trained artificial intelligence to find COVID-19 on ultrasound images of the lungs

Communications Medicine

Scientists have taught artificial intelligence to detect COVID-19 on ultrasound images of the lungs. This mechanism works similarly to recognizing faces in a crowd.

They are the results of their work published in the scientific journal Communications Medicine, informs Medical Xpress.

American scientists trained AI to analyze ultrasound images of the lungs in search of B-lines – bright vertical anomalies that indicate inflammation. It compares computer images with real ultrasound images of patients, with the help of which it recognizes deviations.

“At the beginning of the pandemic, we did not have enough ultrasound images of patients with COVID-19to develop and test our algorithms, and as a result, our deep neural networks never reached peak performance.

We now demonstrate that with computational datasets we can still achieve a high degree of accuracy in estimating and detecting these features COVID-19“, said study co-author Lingyi Zhao.

The use of AI should help hospitals with high volumes patientswhich require quick and accurate diagnosis.

In the future, scientists plan to create wireless devices that patients can use for monitoring progression COVID-19 at home.

“The ideal use case would be wearable ultrasound patches to monitor fluid build-up and tell patients when they need to adjust their medication or see their doctor.” – noted the co-author of the study, Tiffany Fong.

The newly created tool also has the potential to become the basis for the development of devices that monitor diseases such as congestive heart failure, she said.

We will remind, scientists taught AI to diagnose brain tumors on the operating table.