Application of AI Combined with Low-Dose CT Scan in Pulmonary Nodules Screening of
Health Examination Personnel
ZHU Hongwei1, MA Shihua2, KANG Wenjie1, ZHAO Yanan3, CUI Jing1, HU Zhengqian1,
XING Chen1, ZHOU Zhiqiang1, CUI Yijie1
1. Department of Medical Imaging, The No.2 Hospital of Baoding, Baoding Hebei 070151, China;
2. Department of Medical Imaging, Baoding No.1 Hospital, Baoding Hebei 070151, China;
3. Department of Imaging, Baoding Cihui Health Examination Outpatient Department, Baoding Hebei 070151, China
Abstract：Objective To explore the application of artificial intelligence (AI) technology combined with low-dose CT scan in
pulmonary nodules screening of health examination personnel. Methods A total of 350 patients with suspected pulmonary who
underwent health examination in the hospital were enrolled between January 2020 and January 2021. According to different doses of
CT scan, they were divided into observation group (low-dose group, n=175) and control group (routine-dose group, n=175). All data
were analyzed by artificial method and AI technology. They were further divided into routine-dose CT group, low-dose CT group,
AI+routine-dose group and AI+low-dose CT group. Taking pathological results as the diagnostic golden standard, the detection rate
of pulmonary nodules, diagnostic efficiency, detection time of nodules, scan dose, image quality, morphological characteristics of
pulmonary nodules and detection rate of calculi with different diameters among the four groups were compared. Results The gold
standard results showed that the detection rate of pulmonary nodules in observation group and control group were 12.57% (22/175)
and 17.14% (30/175), respectively. There was no significant difference in the detection rate among routine-dose CT group, lowdose
CT group, AI+routine-dose group and AI+low-dose CT group (P>0.05), and the Kappa values were 0.738, 0.722, 0.742 and
0.756, respectively. The reading time of AI was significantly shorter than that of artificial method (P<0.05). The scan radiation
dose parameters, including dose length product, effective tube dose, total tube dose, volume CT dosimetry index in observation
group were significantly lower than those in control group (P<0.05). There was no significant difference in scores of image quality
between the two groups (P>0.05). The difference in pulmonary nodules lobulation, spicule sign, calcification, vocule sign, bronchus
sign and pleural indentation sign between the two groups was not statistically significant (P>0.05). The difference in total detection
rate of pulmonary nodules between observation group and control group was not statistically significant (13.14% vs. 17.71%)
(P>0.05). There was no significant difference in the detection rate of segmentation stones diameter between the two groups (P>0.05).
朱红伟1，马士华2，康文杰1，赵亚楠3，崔靖1，胡政乾1，邢辰1，周志强1，崔铱婕1. AI结合低剂量CT扫描在体检人员肺小结节筛查中的应用[J]. 中国医疗设备, 2022, 37(10): 142-146.
ZHU Hongwei1, MA Shihua2, KANG Wenjie1, ZHAO Yanan3, CUI Jing1, HU Zhengqian1,
XING Chen1, ZHOU Zhiqiang1, CUI Yijie1. Application of AI Combined with Low-Dose CT Scan in Pulmonary Nodules Screening of
Health Examination Personnel. China Medical Devices, 2022, 37(10): 142-146.
