Diagnostic value of lung CT combined with deep learning for pulmonary embolism

ZHAO Xingyong; DI Qun; WEN Huali; MENG Xianli; YANG Jinyong

doi:10.20258/j.cnki.1006-9011.2025.04.010

Abstract：Objective To explore the diagnostic value of pulmonary CT imaging combined with deep learning for pulmonary embolism.Methods A total of 117 patients who underwent pulmonary CT examination and were suspected of having pulmo-nary embolism were selected and divided into the control group(n=117)and the observation group(n=117)based on different diagnostic methods.In the control group,the diagnosis was made by radiologists using conventional pulmonary CT.In the obser-vation group,the diagnosis was assisted by the deep learning model REUNet on the basis of the control group.The deep learning model was built using the popular two-stage object detection technology.The ResNet network was used for feature extraction,and the introduction of residual connections enabled the model to better learn the low-level detail features of CT images,which was particularly important for small target detection.The UNet network was used for object detection,and through the encoding and decoding structure of the network,the CT image information was fully utilized to improve the accuracy and detail retention ability of thrombus detection.The comprehensive diagnostic results of two senior radiologists were taken as the gold standard,and the di-agnostic efficacy and diagnostic time of the two groups were collected and analyzed.Results The gold standard diagnosed 43 positive cases and 74 negative cases.The sensitivity,specificity,accuracy,positive predictive value,and negative predictive value of the observation group were all higher than those of the control group(all P<0.05).The diagnostic time of the observa-tion group was shorter than that of the control group(P<0.05).The area under the curve(AUC)and Kappa value of the observa-tion group were higher than those of the control group,and the differences were statistically significant(all P<0.05).Conclusion The combination of pulmonary CT and the deep learning model REUNet for auxiliary diagnosis can improve the di-agnostic accuracy and efficacy of pulmonary embolism,reduce the diagnostic time,which is conducive to the early and rapid di-agnosis of pulmonary embolism.