Chinese researchers have found that artificial intelligence (AI) can help clinicians distinguish COVID-19 from other pneumonias on chest CT, according to a study published April 27 in Radiology.
Helping clinicians recognize COVID-19 is a key part of effective patient care, wrote a team led by Dr. Harrison Bai of Xiangya Hospital, Central South University, in Changsha, Hunan.
"While distinguishing COVID-19 from normal lung or other lung pathologies such as cancer on chest CT may be straightforward, differentiation between COVID-19 and other pneumonia can be particularly troublesome for physicians because of the radiographic similarities," the group wrote. "Inaccurate imaging interpretation makes it harder for patient management strategies to work efficiently."
Studies have shown that AI can help distinguish COVID-19 from other diseases, but little data exist that compares radiologist performance with and without AI for this purpose.
"Chest CT is often relied on as a supplementary diagnostic measure that helps physicians build a more complete patient assessment," the group wrote. "AI has shown efficacy in differentiating COVID-19 from pneumonia of other etiology on chest CT, yet the practical application of AI augmentation to radiologists' COVID-19 diagnostic workflow has not been explored in the literature."
For their research, Bai and colleagues evaluated whether an AI system could help clinicians distinguish COVID-19 from other diseases and evaluated radiologist performance with and without its help. The study included 1,186 patients, 521 of which were COVID-19 positive on real-time reverse transcription polymerase chain reaction (RT-PCR) testing and chest CT, and 665 of which had non-COVID-19 pneumonia. The total number of cases was divided into training, validation, and test sets for a deep neural network called EfficientNet B4; six radiologists reviewed the studies without AI assistance and then with it.
The AI network performed at a high level on its own and also improved the performance of radiologists, the group found. All results were statistically significant.
| AI's effect on distinguishing COVID-19 from non-COVID-19 pneumonia on chest CT | |||
| Performance measure | Radiologist alone | AI alone | Radiologist with AI assistance |
| Accuracy | 85% | 96% | 90% |
| Sensitivity | 79% | 95% | 88% |
| Specificity | 88% | 96% | 91% |
"Our study revealed that when compared to a radiologist-only approach, AI augmentation significantly improved radiologists' performance distinguishing COVID-19 from pneumonia of other etiology, yielding higher measures of accuracy, sensitivity, and specificity," the authors wrote.
Although in general, the diagnostic accuracy of COVID-19 chest CT interpretation is good, it needs to be better in order to best manage both healthcare facilities' resources and patient care during the pandemic, Bai's group noted. That's why the study results are promising -- they indicate that incorporating AI into radiologists' chest CT workflow could improve COVID-19 diagnosis.
"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," the authors concluded. "The results that we present suggest that integrating AI into radiologists' routine workflow has potential to improve diagnostic outcomes related to COVID-19."
