Abstract:Objective To study the diagnostic value of artificial intelligence (AI) technology based on deep learning in different
classification of rib fractures. Methods The CT images of 77 patients with rib fracture caused by chest trauma were analyzed
retrospectively. First, two junior radiologists and uAI EasyTriage-Rib software (Shanghai United Imaging Medical Co., Ltd.) read the
images separately, and then the radiologists used uAI EasyTriage-Rib software read the images again. The fracture point and fracture
type were recorded, together with the diagnosis time. The detection rate, detection accuracy and sensitivity of different types of rib
fractures and the false positive and false negative in AI diagnosis were analyzed. Results The detection rate of uAI EasyTriage-
Rib software for different types of rib fractures were 96.04% (dislocation fracture), 97.02% (non-dislocation fracture), 98.79% (old
fracture). The diagnostic accuracy and sensitivity of uAI EasyTriage-Rib software for different types of rib fractures were higher than
those of low-seniority radiologists (P<0.05). The diagnostic accuracy of radiologists using AI for different types of rib fractures was
higher than that of AI alone (P<0.05), and the reading time of using uAI EasyTriage-Rib software was significantly lower than that of
radiologists alone (P<0.05). The false rate in the diagnosis of all types of rib fractures with uAI EasyTriage-Rib software were 2.53%.
Conclusion AI can improve the diagnostic efficiency of radiologists effectively for different types of rib fractures and has great
clinical application value. Joint diagnosis by physician and AI is more effective.
李星宇,雷禹,黄晓旗,李建龙,闫军. 基于AI与人工阅片对肋骨骨折性质诊断价值的比较[J]. 中国医疗设备, 2022, 37(8): 32-36.
LI Xingyu, LEI Yu, HUANG Xiaoqi, LI Jianlong, YAN Jun. Comparison of the Diagnostic Value of Rib Fracture Properties Based on AI and Manual Reading. China Medical Devices, 2022, 37(8): 32-36.