Artificial Intelligence Detects Undiagnosed Liver Disease from Echocardiograms

This imaging test is commonly used as an early screening method when doctors suspect a patient may have cardiovascular disease. A typical echocardiogram may contain more than 50 video clips, with some showing images of the liver. People with heart disease often develop chronic liver conditions, making it difficult to differentiate between primary liver disease and liver injury resulting from heart disease. It’s estimated that 4.5 million individuals have been diagnosed with liver disease, although many more may have undiagnosed steatotic liver disease, previously known as fatty liver disease. Now, a new artificial intelligence (AI) program can detect chronic liver disease from videos captured during an echocardiogram. This deep-learning model aids doctors in identifying liver disease that might otherwise go unnoticed, leading to timely follow-up testing.

Researchers at Cedars-Sinai (Los Angeles, CA, USA) trained an AI system to analyze patterns in over 1.5 million echocardiogram videos. The AI program, called EchoNet-Liver, was able to detect cirrhosis (scarring of the liver) and steatotic liver disease by examining liver images captured during echocardiograms. This technology builds upon the earlier-developed EchoNet, which identifies and analyzes patterns in echocardiogram images. The team compared the deep-learning model's predictions to diagnoses made using abdominal ultrasounds or MRI images. According to a study published in NEJM AI, the AI model's accuracy was comparable to that of traditional imaging methods analyzed by radiologists. The next phase of research involves testing EchoNet-Liver in studies that track patients' health over time.

“These novel findings exemplify how AI models are helping us to augment clinical diagnostics at a body-systems level instead of just individual organs,” said Sumeet Chugh, MD, director of the Division of Artificial Intelligence in Medicine at Cedars-Sinai.