X-rays, first used clinically in the late 1890s, could be a pioneering diagnostic tool for COVID-19 patients with the help of artificial intelligence, according to a team of researchers in Brazil who taught a computer program, through various machine learning. Methods, to detect COVID-19 in chest X-ray with 95.6 to 98.5% accuracy.
They published their results in IEEE / CAA Journal of Automatica Sinica, Which is a joint publication between IEEE and China Automation Association.
Researchers have previously focused on detecting and classifying lung diseases, such as fibrosis, emphysema, and lung nodules, through medical imaging. Common symptoms of suspected COVID-19 infection include shortness of breath, coughing and pneumonia in more severe cases – all of which are visible on medical imaging such as a CT scan or X-ray.
“When the COVID-19 pandemic emerged, we agreed to use our experience to help deal with this new global problem,” said corresponding author Victor Hugo C.D. Albuquerque, researcher at the Laboratory for Image, Signal and Application Processing. Computing and with the University of Fortaleza.
Albuquerque said many medical facilities have an insufficient number of tests and long treatment times, so the research team has focused on improving a tool readily available in every hospital and already used frequently in diagnosing COVID-19: X-ray machines.
Albuquerque said, “We decided to investigate whether COVID-19 infection could be detected automatically using X-ray images,” noting that most X-ray images are available within minutes, compared to the days required for diagnostic swab or saliva tests.
However, researchers found a shortage of publicly available X-rays to train their AI model to automatically recognize the lungs of COVID-19 patients. They had 194 COVID-19 mammograms and 194 healthy mammograms, while it usually takes thousands of images to teach a comprehensive model for detection and classification of a specific target. To compensate, they took a model that was trained in a large dataset from other X-ray images and trained it to use the same techniques to detect lungs potentially infected with COVID-19. They used several different methods of machine learning, two of which resulted in an accuracy rate of 95.6% and 98.5%, respectively.
“Because X-rays are so fast and cheap, they can help triage patients in places where the healthcare system has collapsed or in locations far from major centers with access to more sophisticated technologies,” Albuquerque said. “This approach to automatically detecting and classifying medical images can assist clinicians in identifying and measuring the severity and classification of disease.”
After that, Albuquerque said, the researchers plan to continue testing their method with larger data sets as they become available, with the ultimate goal of developing a free online platform for classifying medical images.
EF Ohata, GM Bezerra, JVS Chagas, AV Lira Neto, AB Albuquerque, VHC Albuquerque, and PP Rebouças Filho, “Automatic detection of COVID-19 infection using chest X-ray images through learning transfer” IEEE / CAA J. Autom. Seneca, Vol. 8, no. 1, p. 239-248, Jan 2021.
Full text available:
IEEE / CAA Journal of Automatica Sinica It aims to spread high-quality, useful and far-reaching research achievements globally, providing an international forum for presenting original ideas and recent findings related to all aspects of automation.
Impact factor I IEEE / CAA Journal of Automatica Sinica It is 5,129, and it ranks among the top 17% (11/63, SCI Q1) in the automation and control systems category, according to the latest journal citation reports released by Clarivate Analytics in 2020. In addition, the latest release of CiteScore is 8.3, and contains Q1 entered in all three categories to which it belongs (Information System, Control and Systems Engineering, Artificial Intelligence) since 2018.
Why post with us: Fast, high-quality peer review; Simple and efficient online submission system; The widest possible global reach of your research; Indexed in SCIE, EI, IEEE, Scopus, Inspec. JAS papers can be found at http: // ieeexplore.