AI helps radiologists distinguish COVID-19 from conventional pneumonia
A new AI model can help radiologists distinguish COVID-19 from non-COVID 19 pneumonia when reading chest CT examinations. The researchers behind this new algorithm shared their findings in Radiology.
“It has been hypothesized that COVID-19 infection is difficult to contain because of its potential transmission from asymptomatic carriers,” wrote lead author Harrison X. Bai, MD, Rhode Island Hospital in Providence, and colleagues. “Common symptoms include fever, cough, and dyspnea while the disease has potential to cause a host of severe and potentially fatal cardiorespiratory complications in vulnerable populations—particularly the elderly with comorbid conditions. While distinguishing COVID-19 from normal lung or other lung pathologies such as cancer on chest CT may be straightforward, a major hurdle in controlling the current pandemic is making out subtle radiological differences between COVID-19 and pneumonia of other etiology.”
Bai et al. gathered imaging data from 521 COVID-19 patients treated in China or the United States from January to April 2020. Findings from 665 pneumonia patients who did not have COVID-19 were also collected; those patients were treated in China or the United States from Jan. 1, 2017, to Dec. 31, 2019.
The team’s deep learning model was trained to distinguish COVID-19 imaging results from pneumonia results. The training set included 830 patients, the validation set included another 237 patients and the test set included 119 patients. Six blinded radiologists were then brought in to evaluate the examinations with—and then without—AI assistance.
Overall, the AI model had a higher test accuracy (96% vs 85%), sensitivity (95% vs 79%) and specificity (96% vs 88%) than radiologists alone. When radiologists used the AI model to help guide their diagnosis, their accuracy (90% vs 85%), sensitivity (88% vs 79%) and specificity (91% vs 88%) all saw notable improvements.
“Our study is relevant and novel for demonstrating the effect of AI augmentation on radiologist performance in distinguishing COVID-19 from pneumonia of other etiology on chest CT,” Bai and colleagues wrote. “The results that we present suggest that integrating AI into radiologists’ routine workflow has potential to improve diagnostic outcomes related to COVID-19.”
Michael has more than 18 years of experience as a professional writer and editor. He has written at length about cardiology, radiology, artificial intelligence and other key healthcare topics.