Article image

AI can detect COVID-19 from ultrasound images with incredible accuracy

Artificial intelligence (AI) has shown the capability to identify COVID-19 from lung ultrasound images with a precision likened to how facial recognition technology can pinpoint a face in a crowded space.

This advancement signifies a significant leap in AI-driven medical diagnostics. Additionally, it brings the medical community closer to achieving quick and accurate diagnoses of COVID-19 and other pulmonary diseases.

This is made possible through the use of sophisticated algorithms that analyze ultrasound imagery for disease indicators.

Assisting overworked doctors

The study, recently published in Communications Medicine, marks the culmination of efforts initiated early in the COVID-19 pandemic. During this time, clinicians urgently needed efficient tools to quickly assess numerous patients in crowded emergency departments.

Muyinatu Bell, the study’s senior author and Johns Hopkins University’s John C. Malone Associate Professor, introduced the tool.

“We developed this automated detection tool to assist doctors in emergency settings with high caseloads of patients who require prompt and precise diagnoses, akin to the earlier stages of the pandemic,” Bell stated.

She also highlighted the goal of allowing patients to monitor COVID-19 progression at home with wireless devices.

AI, COVID-19, and the future of wearable health tech

Tiffany Fong, an assistant professor of emergency medicine at Johns Hopkins Medicine and co-author of the study, discussed the technology’s potential.

She highlighted its application in developing wearables for diseases like congestive heart failure, which shares COVID-19 symptoms such as fluid accumulation in the lungs.

Fong envisions, “An ideal use case would be wearable ultrasound patches that monitor fluid buildup and alert patients when they need a medication adjustment or a doctor’s visit.”

How this AI tool detects COVID-19

The AI tool operates by analyzing ultrasound images of the lungs to identify features known as B-lines. These features, appearing as bright, vertical lines, signal inflammation commonly seen in patients with pulmonary issues.

To achieve accurate detection, the AI merges computer-generated images with actual patient ultrasounds. This includes images from individuals treated at Johns Hopkins.

Bell explained the complex process, stating, “We had to accurately model the physics of ultrasound and acoustic wave propagation to create believable simulated images. Subsequently, we trained our computer models to interpret real scans from patients with lung complications using these simulated data.”

Overcoming early obstacles

The journey to this breakthrough encountered significant obstacles. Initially, the scarcity of patient data and a limited understanding of COVID-19’s effects posed a challenge to the AI’s diagnostic accuracy.

However, by developing software that leverages both real and simulated data, Bell’s team enhanced the AI’s ability to identify signs of COVID-19 in ultrasound scans.

This improvement is attributable to deep neural networks. This type of AI mimics the brain’s pattern recognition capabilities, resulting in a significant performance boost.

Lingyi Zhao, the study’s first author and a former member of Bell’s team, is now with Novateur Research Solutions.

She observed, “The early shortage of COVID-19 ultrasound images limited our development and testing. But with computer-generated datasets, we’ve achieved a high accuracy level in detecting COVID-19 features.”

Implications and future study of AI and COVID-19 diagnosis

This incredible technological advancement demonstrates the powerful capabilities of artificial intelligence (AI) in transforming medical diagnostics, specifically in the rapid and accurate detection of COVID-19 and other pulmonary diseases through lung ultrasound images.

By harnessing the potential of AI-driven tools and developing innovative solutions like wearable devices for continuous monitoring, researchers at Johns Hopkins University pave the way for a future where healthcare professionals can diagnose and manage diseases more efficiently.

This advancement streamlines the diagnostic process in emergency settings and offers promising avenues for patient self-monitoring, highlighting the significant strides made in integrating technology with healthcare to improve patient outcomes.

The full study was published in the journal Communications Medicine.


Like what you read? Subscribe to our newsletter for engaging articles, exclusive content, and the latest updates. 

Check us out on EarthSnap, a free app brought to you by Eric Ralls and


News coming your way
The biggest news about our planet delivered to you each day