Deep-Learning Assessment of Liver Fat Identifies Risk of Severe COVID-19

Researchers from Emory University and University Hospitals (UH) Cleveland have developed a deep learning (DL)-based liver fat assessment tool which shows that patients with nonalcoholic fatty liver disease, also known as hepatic steatosis, are one-and-a-half times more likely to develop severe COVID-19.

The tool, a deep learning-based hepatic fat assessment (DeHFt) pipeline, is designed to provide automated measurements of liver fat from standard computerized tomography (CT) scans. According to the press release, using CT scans to determine whether a patient has fatty liver disease typically relies on a time-consuming manual examination of scans. Using DeHFt has the potential to allow clinicians to speed up this process.

However, a recent study shows that the tool also has applications for risk stratification and disease severity assessment for various conditions, including COVID-19.

“We know that hepatic steatosis is a risk factor for COVID-19. Now we can use this pipeline to identify high-risk patients and based upon that clinicians can make better informed decisions about levels of care and the early use of therapeutics, such as antivirals,” said Gourav Modanwal, PhD, first author of the study and a researcher in the Wallace H. Coulter Department of Biomedical Engineering at Emory University School of Medicine and Georgia Institute of Technology College of Engineering, in the press release.

DeHFt relies on coronary artery calcium CT scans, which are commonly used to detect and measure calcium-containing plaque in patients' arteries. The images from these scans show the heart and portions of the spleen and liver, which provide an opportunity to evaluate patients' liver fat.

Despite this, the press release notes that these scans have not historically been used to assess hepatic fat because of difficulties in manually measuring regions of interest at higher resolutions and magnifications.

The DeHFt pipeline uses DL to perform these assessments. First, a segmentation model is trained to segment the liver and spleen using coronary artery calcium CT scans. Then, the algorithm estimates CT attenuation, a measure of density, of the liver and spleen using stacks of liver and spleen slices visible in 3D.

In this case, lower CT attenuation values are associated with more fat infiltration in the liver, while the spleen serves as a control to compare the liver-to-spleen ratio.

According to the study, published in The Lancet’s eBioMedicine open-access journal, the DeHFt pipeline accurately segmented the liver and spleen on non-contrast CTs and automated the estimation of liver and liver-to-spleen attenuation ratio. Researchers also found that hepatic steatosis was linked to disease severity among COVID-19 patients.

“This is a very exciting and translationally relevant finding. Our study suggests that machine learning based on routine CT scans can help in accurate quantification of liver fat, which has implications that extend beyond COVID-19 severity assessment,” says Anant Madabhushi, PhD, the study’s senior author and a professor in the Wallace H. Coulter Department of Biomedical Engineering at Emory School of Medicine and Georgia Institute of Technology College of Engineering, in the press release.

These attenuation values also serve as surrogate markers for cardiometabolic risks, such as type 2 diabetes and its progression, according to the press release.

“This novel pipeline provides an important avenue for CT-based analysis of adiposity and metabolic risk that is scalable for population-level imaging and can be used for risk stratification for cardiometabolic disease,” said study co-author Sadeer Al-Kindi, MD, director of the cardiovascular phenomics core and co-director of the Vascular Metabolic Center at the UH Harrington Heart & Vascular Institute, in the press release. “We are currently working on translating this and validating it in various large cohorts for risk prediction.”

The researchers concluded that though DeHFt needs to be further validated in other studies, it has the potential to provide reliable, reproducible liver fat measurements, which may lead to an integrated cardiometabolic and COVID-19 risk tool.