Abstract:Objective To investigate the efficacy of artificial intelligence (AI) technology in chest high resolution CT (HRCT) in the diagnosis of pulmonary nodules. Methods The data of 130 patients with single pulmonary nodule received chest HRCT in Shenyang Hospital of Traditional Chinese Medicine from January 2020 to January 2021 were analyzed retrospectively. All patients were diagnosed by surgery and pathology. The characteristics of nodules were compared and analyzed by AI image analysis system and manual reading, including the size, density and risk of nodules. The detection time, missed diagnosis rate, false positive rate and accuracy of AI and artificial were counted. Results The time of AI detecting pulmonary nodules was (73.10±28.35) s, the missed diagnosis rate was 1.54%, the false positive rate was 23.35%, and the coincidence rate with pathological diagnosis was 72.31%. The time of manual reading of pulmonary nodules was (273.26±68.62) s, the missed diagnosis rate was 6.92%, the false positive rate was 0, and the coincidence rate with pathological diagnosis was 90.00%. The detection time and detection rate of AI in pulmonary nodules were significantly better than those of manual reading, and the false positive rate and coincidence rate with pathological diagnosis of manual reading were better than those of AI (P<0.05). Conclusion Although the false positive rate of AI technology is high, and there is still a certain gap between AI technology and manual reading in the qualitative analysis of nodules, still AI could assist manual reading, and has a significant effect in improving work efficiency and reducing missed diagnosis rate.