A Study Of The Application Of Radial Ultrasound Diagnostic Modality In Peripheral Pulmonary Lesions

Objective:We retrospectively analyzed the differences in radial endobronchial ultrasound images of peripheral pulmonary lesions to find the endobronchial ultrasound image features that predict benign and malignant lesions in the peripheral lung and constructed a predictive model of benign and malignant lesions by combining high-resolution CT to provide a diagnostic basis for determining benign and malignant peripheral pulmonary lesions.Methods:This study is a retrospective study.186 patients who underwent ultrasound bronchoscopy had definite diagnosis at the First Affiliated Hospital of Kunming Medical University from August 2019 to November 2022.We focused on R-EBUS image features and HRCT imaging features and revealed the most meaningful signs for the differential diagnosis of two types of lesions by univariate factor and multivariate logistic regression analysis.The prediction models for the benignity and malignancy of peripheral pulmonary lesions by HRCT,R-EBUS,and HRCT combined with R-EBUS were established and validated for their predictive efficacy respectively.Results:1.There were statistical differences in EBUS image features(shape,margin,anechoic zone,linear bronchial inflation sign,hypoechoic zone)and HRCT imaging features(lobar sign,burr sign,pleural indentation sign)between the two groups for benign and malignant lesions.2.Between the two groups of benign and malignant lesions,R-EBUS multivariate logistic regression analysis showed that four predictors,including shape,margin,hypoechoic area,and linear bronchial inflation sign,were relatively independent risk factors for determining the benignity and malignancy of PPLs;HRCT multivariate logistic regression analysis showed that two predictors,including lobar sign and pleural indentation sign,were relatively independent risk factors for determining the benignity and malignancy of PPLs.The R-EBUS combined with HRCT multivariate logistic regression analysis showed that five predictors,including shape,margin,hypoechoic area,lobar sign and pleural indentation sign,were relatively independent risk factors for determining the benignity and malignancy of PPLs.3.In the prediction model of PPLs,EBUS benign and malignant lesion prediction model:P=e~x/(1+e~x),x=-7.812+(1.387×shape)+(1.053×margin)-(1.046×linear bronchial inflation sign)+(3.242×hypoechoic area);HRCT benign and malignant lesion prediction model:P=e~x/(1+e~x),x=-1.643+(1.106×lobar sign)+(1.663×pleural indentation sign);R-EBUS combined with HRCT benign and malignant lesion prediction model:P=e~x/(1+e~x),x=-9.960+(1.583×shape)+(0.851×margin)+(3.610×hypoechoic area)+(1.436×lobar sign)+(1.592×pleural indentation sign).4.In the judgment of model prediction effect,the sensitivity of R-EBUS benign and malignant lesion prediction model was 82.1%and the specificity was 89.9%;the sensitivity of HRCT benign and malignant lesion prediction model was 68.6%and the specificity was 76.3%;the sensitivity of R-EBUS combined with HRCT benign and malignant lesion prediction model was 88.1%and the specificity was 93.3%.Conclusions:1.R-EBUS image features can assist in predicting the benignity and malignancy of peripheral pulmonary lesions,which has a high application value.Malignant lesions often show lobar shape,distinct but not sharp margins,the presence of anechoic areas,the absence of linear bronchial inflation signs,and the presence of hypoechoic areas,while benign lesions have no characteristic performance.2.The prediction model established by combining R-EBUS features and HRCT features has higher sensitivity and specificity than prediction model of R-EBUS or prediction model of HRCT,which has certain reference significance for clinical purposes.