Artificial Intelligence-Based Detection of Pneumonia in Chest Radiographs

Abstract

Artificial intelligence is gaining increasing relevance in the field of radiology. This study retrospectively evaluates how a commercially available deep learning algorithm can detect pneumonia in chest radiographs (CR) in emergency departments. The chest radiographs of 948 patients with dyspnea between 3 February and 8 May 2020, as well as 15 October and 15 December 2020, were used. A deep learning algorithm was used to identify opacifications associated with pneumonia, and the performance was evaluated by using ROC analysis, sensitivity, specificity, PPV and NPV. Two radiologists assessed all enrolled images for pulmonal infection patterns as the reference standard. If consolidations or opacifications were present, the radiologists classified the pulmonal findings regarding a possible COVID-19 infection because of the ongoing pandemic. The AUROC value of the deep learning algorithm reached 0.923 when detecting pneumonia in chest radiographs with a sensitivity of 95.4%, specificity of 66.0%, PPV of 80.2% and NPV of 90.8%. The detection of COVID-19 pneumonia in CR by radiologists was achieved with a sensitivity of 50.6% and a specificity of 73%. The deep learning algorithm proved to be an excellent tool for detecting pneumonia in chest radiographs. Thus, the assessment of suspicious chest radiographs can be purposefully supported, shortening the turnaround time for reporting relevant findings and aiding early triage.

Clinical Implication and Prognostic Value of Artificial-Intelligence-Based Results of Chest Radiographs for Assessing Clinical Outcomes of COVID-19 Patients

“…There have been many recent efforts to use AI for diagnosing and validating prognostic power for management of pulmonary infection. Several studies have already been conducted to find the meaning of AI on CXR in diagnosing pneumonia, evaluating treatment response, or predicting prognosis [ 18 , 19 , 20 ]. In addition to pneumonia, researchers have focused on developing an AI algorithm to detect COVID-19 using CXR because CXR is the first-line imaging study for diagnosing and guiding treatment options of patients with respiratory symptoms [ 21 , 22 , 23 , 24 ].…”

Utilizing Different Deep Learning Models to Accurately Detect Pneumonia in Chest X-ray Images

“…However, this is not to say that AI cannot aid in logically and arithmetically processing data in ways that could succor current health fields. In one study, an AI model did remarkably well, successfully identifying patients with pneumonia as suspected to have COVID-19 with an accuracy of 95.4% [11]. AI even outperformed humans in identifying breast cancer in ultrasounds, with the average of 10 radiologists being 0.924 ± 0.02 and the model having an AUROC of 0.962.…”

The study on the predictive accuracy of artificial intelligence (AI) Lunit INSIGHT CXR Version 3.0 for pneumonia diagnosis in COVID-19 patients

“…Due to the limited number of radiologists, it may not be possible to make a timely diagnosis if there are many patients, such as the recent COVID-19 infection. CXRs are the most popularly performed imaging tests; yet, a rapid Discussion interpretation of CXRs by radiologists is problematic in hospitals, especially for those with severe lesions [22] and rapid identification of lung infections may result in speedier isolation of patients, potentially reducing the risk of infection spread [23]. Recently, many studies have demonstrated the potential for AI application in radiology as the clinical decision support system (CDSS) or even as a second reading [24] and has great potential for the analysis of large amounts of data.…”