Clinical Characteristics Ultrasound Misdiagnosis of Breast Cancer and Construction of Risk Prediction Model

CHEN Jing

doi:10.3969/j.issn.1002-3429.2024.11.005

Abstract：Objective To analyze the clinical characteristics of ultrasound breast cancer misdiagnosed by ultrasound and to construct a risk prediction model for misdiagnosis.Methods A total of 602 patients with breast mass who underwent ultrasound examination from January 2021 to December 2023 were selected and divided into misdiagnosis group and accurate diagnosis group according to the presence of misdiagnosis.The characteristics of patients misdiagnosed were analyzed,and a risk prediction model for misdiagnosis was constructed.Results A total of 30 patients were misdiagnosed as breast cancer,and the misdiagnosis rate was 4.98%.There were significant differences between the accurate diagnosis group and the misdi-agnosis group with respect to the largest diameter of the lesion,palpation,combined benign lesion,combined inflammatory le-sion,boundary of ultrasound image,obvious blood flow signal,axillary lymph node echo,microcalcification,and complex and diverse lesion structure(P<0.01).Binary Logistic regression analysis showed that the largest diameter of the lesions(≥5 cm),combined benign lesions,combined inflammatory lesion,boundary of ultrasound image(unclear),obvious blood flow signal,axillary lymph node echo,microcalcification,and complex and diverse lesion structure were risk factors for ultrasound misdi-agnosis of breast cancer(P<0.05,P<0.01).Risk prediction model of ultrasound misdiagnosis of breast cancer was as follows:2.623×the maxi-mum diameter of lesion(≥5 cm)+1.422×benign lesion+1.616×inflammatory lesion+1.574×boundary of ultrasound image(blurred)+1.134×obvious blood flow signal+1.518×axillary lymph node echo+2.027×microcalcifica-tion+1.541×complex and diverse lesion structure.Hosmer-Lemeshow fit test showed that the model had a good fit(χ2=3.487,P=0.900).The analysis of receiver operating characteristic(ROC)curve showed that the area under the ROC curve(AUC)of the prediction model for preciting ultrasound misdiagnosis of breast cancer was 0.921,the Yodon index was 0.698,and the sensitivity and specificity were 83.3%and 86.5%,respectively;the 95%CI was 0.872 and 0.969,and the actual appli-cation accuracy was 96.2%(579/602).Conclusion Ultrasound misdiagnosis of breast cancer is related to a variety of factors.The risk prediction model of ultrasound misdiagnosis of breast cancer based on the maximum diameter of lesion,benign lesion,inflammatory lesion,unclear boundary of ultrasound image,obvious blood flow signal,axillary lymph node echo,microcalcifi-cation and complex lesion structure has good prediction accuracy.