AI Model Can Diagnose Fatty Liver Disease Using Chest X-Ray: Study

Researchers from Osaka Metropolitan University have created an AI-based model capable of identifying the condition using routine chest X-rays. The study, published in the journal Radiology Cardiothoracic Imaging, demonstrates that the new method offers a cost-effective and widely accessible alternative to current diagnostic tools that rely on specialized imaging equipment.

Fatty liver disease affects about one in four people worldwide and can lead to cirrhosis or liver cancer if untreated. While standard diagnostic tools like ultrasounds, CT scans, and MRIs are costly and require specialized equipment, chest X-rays are cheaper, widely used, and expose patients to minimal radiation. Though chest X-rays capture part of the liver, their potential for detecting fatty liver disease has been largely unexplored.

Recognizing this gap, Associate Professor Sawako Uchida-Kobayashi and Associate Professor Daiju Ueda from the Graduate School of Medicine at Osaka Metropolitan University led a study to develop an artificial intelligence model capable of detecting fatty liver disease from chest X-ray images. Using a retrospective dataset of 6,599 chest X-rays from 4,414 patients, the team trained the AI model with controlled attenuation parameter (CAP) scores, a recognized metric for liver fat content.

The AI model demonstrated high diagnostic accuracy, with the area under the receiver operating characteristic curve (AUC) ranging from 0.82 to 0.83, indicating strong predictive performance.

“The development of diagnostic methods using easily obtainable and inexpensive chest X-rays has the potential to improve fatty liver detection. We hope it can be put into practical use in the future,” said Professor Uchida-Kobayashi.

This AI-driven approach could significantly enhance early diagnosis and intervention for fatty liver disease, particularly in low-resource settings where advanced imaging tools are not readily available. By leveraging common medical imaging and artificial intelligence, the study paves the way for more accessible liver health assessments worldwide.

Reference: Ueda, D., Uchida-Kobayashi, S., Yamamoto, A., Walston, S. L., Motoyama, H., Fujii, H., ... & Kawada, N. (2025). Performance of a Chest Radiograph–based Deep Learning Model for Detecting Hepatic Steatosis. Radiology: Cardiothoracic Imaging, 7(3), e240402.