Can AI based on chest x-rays predict COVID-19 mortality risk?

A group led by Yoshikazu Uchiyama, PhD, from Kumamoto University in Japan, developed an AI model trained on the portable X-ray features of COVID-19 and found that it performed well at identifying deaths versus recoveries. The results indicate that X-rays can provide predictive data that are normally obtained from computed tomography.

“In previous radiological studies of COVID-19, computed tomography (CT) images were used due to ease of analysis, while chest radiographs were not,” the group wrote.

The researchers extracted portable chest X-ray images of 100 patients from the Cancer Imaging Archives’ COVID-19 database, a service provided by the US National Institutes of Health. Of these individuals 10 died and 90 recovered. Their goal was to develop a model that distinguishes between these states.

Also Read :  Get Ready to Relearn How to Use the Internet

Because portable chest X-ray images of patients with COVID-19 can be difficult to interpret — because bone components overlap with abnormal patterns of disease — the group used bone suppression technology while pre-processing the images, testing a model with and without it.

The team then included 620 radiological features in the model based on measurements in the left and right lung regions on X-rays, including a volume feature, nine histogram features, 272 histological features, and 28 resolution features. The group analyzed the model’s performance using the area under the receiver operating characteristic (AUC) curve.

Also Read :  AI solved a debate on who made this dinosaur footprint

Examples of images with osteotomy and manually segmented lung regions

Examples of images with osteotomy and manually segmented lung regions. Courtesy of Journal of Medical Radiation Sciences.

To distinguish between death and recovery cases, the highest AUCs achieved by the model were 0.756 and 0.959, depending on the number of radiomic features included. But its discriminatory performance improved when the researchers used a bone suppression technique with a sensitivity of 90.9% and specificity of 95.6% to predict mortality or survival.

Also Read :  Turning A Passion Into A Boutique Travel Business: Don’t Think, Just Do

They noted that an AI approach that uses portable chest X-rays instead of CT scans to predict the outcome of COVID-19 disease could reduce the risk of infection among hospital staff, as images can be obtained in patient rooms. But more research is needed.

The group concluded, “In the future, we will explore the utility of using portable chest radiographs for this purpose by increasing the number of cases and improving the reliability of the system.”

Copyright © 2022


Leave a Reply

Your email address will not be published. Required fields are marked *

Related Articles

Back to top button