Skip to content
Revolutionary AI Now Able to Detect COVID-19 in Lung Ultrasound Images

Revolutionary AI Now Able to Detect COVID-19 in Lung Ultrasound Images

Artificial intelligence is once again providing revolutionary solutions in the fight against COVID-19. A team of researchers from the esteemed Johns Hopkins University has confirmed that AI can now efficiently detect signs of COVID-19 in lung ultrasound images - a breakthrough discovery set to streamline the diagnostic process immensely.

Research findings, recently published in Communications Medicine, show a striking resemblance in how AI can precisely identify a face in a crowded place and the way it can now detect COVID-19 symptoms in ultrasound lung images. This advancement is set to significantly enhance the diagnostic efficiency of health care professionals dealing with COVID-19 and other pulmonary diseases.

This achievement is the culmination of a tireless effort sparked at the onset of the global pandemic, a time when clinicians desperately needed rapid assessment tools to manage an influx of COVID-19 patients.

The team, led by senior author Muyinatu Bell, from Johns Hopkins University explained "The tool was developed to assist with swift and accurate diagnosis in high caseload emergency settings, such as those experienced during the initial stages of the pandemic."

Moreover, the team envisions using this AI-driven detection tool to monitor the progression of the virus in patients' lungs remotely. This can potentially be achieved by integrating it into wireless devices that patients can use at home.

The innovative tool also holds potential for the development of wearables that could allow for the monitoring of other illnesses, such as congestive heart failure, which is known to cause fluid overload in patients' lungs, similar to COVID-19.

The team's AI system works by examining ultrasound lung images to identify features known as B-lines—bright, vertical abnormalities indicative of inflammation in patients with pulmonary complications. The AI software combines simulated images and real patient ultrasounds—including those from patients treated at Johns Hopkins—to detect these abnormalities.

Despite facing struggles at the onset of the pandemic due to a lack of patient data and understanding of COVID-19 symptoms manifestation, the team was able to develop a deep neural network tool. This AI tool, designed to mimic a human brain's interconnected neurons, can recognize patterns and interpret complex tasks. This includes learning from both real and simulated data to detect COVID-19 abnormalities in ultrasound scans.

Lead scientist Lingyi Zhao also noted that "Although we didn't initially have enough ultrasound images of COVID-19 patients for our algorithms to reach peak performance, we proved that with computer-generated datasets, our deep neural network still achieved a high degree of accuracy in evaluating and detecting these COVID-19 features."

With such advancements in AI, the future of healthcare diagnosis and patient monitoring is promising. As the diagnostic AI tools continue developing, the day might not be far when wearable ultrasound patches and home-monitoring devices become the norm, leading to a more proactive, patient-driven healthcare approach.

Disclaimer: The above article was written with the assistance of AI. The original sources can be found on ScienceDaily.