Diagnosis of lung nodule and lung cancer on screening chest radiographs: comparative clinical trial for evaluation of artificial intelligence-integrated PACS versus conventional PACS

INTERVENTION: Medical Device : In case group, chest radiograph interpretation will be performed using a deep learning‐based computer‐aided detection software (Software name: Lunit INSIGHT CXR MCA, Manufacturer: Lunit Inc.). The software was approved as a medical device by the Ministry of Food and Drug Safety, Republic of Korea. The software was designed to analyze a chest radiograph image, and identify relevant abnormalities (nodule, consolidation, and pneumothorax) on the image, and to assist the interpretation by doctors. The software provides a probability value for the presence of abnormality on the chest radiograph, and localization of abnormality overlaid on the original image. The duty board‐certified radiologist will interpret chest radiographs after evaluation of both original chest radiograph image and the result from the computer‐aided detection system. In control group, the duty board‐certified radiologist will interpret chest radiographs after evaluation of original chest radiograph images only, as conventional clinical practice. CONDITION: Diseases of th respiratory system PRIMARY OUTCOME: Lung nodule detection rate SECONDARY OUTCOME: CT referral rate Detection rate of other lung disease False referral rate Lung cancer detection rate Negative predictive value Positive predictive value Positive rate Revisit rate within three months Sensitivity Specificity INCLUSION CRITERIA: All adult individuals who performed chest radiograph in Seoul National University Healthcare Screening Center