Research on the Value of Artificial Intelligence Chest CT Lung Nodule Detection Based on Deep Learning

Abstract

Objective to explore the value of artificial intelligence chest CT lung nodule detection based on deep learning.
Methods a total of 300 patients who underwent chest CT scans in our hospital from January 2019 to October 2020 were
collected as the research objects, and the lung nodule artificial intelligence recognition system was introduced to automatically
identify lung nodules, and the original impact detection results were carried out. To compare the location, size and
characteristics of lung nodules detected by two chief chest imaging physicians, compare the detection rate of lung nodules by
artificial intelligence software and imaging physicians, and the average time spent on the imaging. Results after confirmation by
2 chief physicians, 1123 true nodules were detected, with a diameter of 0.3~2.8cm, of which 158 were malignant nodules; the
detection rate of small and medium nodules based on artificial intelligence software based on deep learning significantly higher
than the detection rate of nodules by imaging physicians, and the detection rate of solid nodules and ground glass density nodules
is significantly higher than that of imaging physicians, subpleural nodules, peripheral nodules, and central nodules etc. The detection
rate of lung nodules at the location was higher than that of the imaging physician (P＜0.05); the artificial intelligence reading time
was (0.1±0.0) min, which was significantly shorter than that of the imaging physician (5.12±1.15) min (P＜0.05). Conclusion the
artificial intelligence system based on deep learning can detect malignant pulmonary nodules in a short time, which is higher than that
of physicians in detecting pulmonary nodules. It can be used as an effective auxiliary tool for the diagnosis of lung nodules.