Establishment of A Model for Diagnosing Lung Adenocarcinoma in Situ using Ultra-high Resolution CT Features
Objective To establish a prediction model of clinical outcome based on ultra-high resolution CT signs in the diagnosis of lung adenocarcinoma in situ(AIS),and explore the predictive value of ultra-high resolution CT signs in the diagnosis of AIS.Methods A total of 110 AIS patients treated in our hospital from January 2019 to December 2022 were selected as research objects.All patients received CT scanning and standardized comprehensive treatment after admission,and ultra-high resolution CT signs were collected.Besides,15-day clinical outcomes were statistically analyzed,and they were divided into outcome group and adverse group.The relationship between ultra-high resolution CT signs and AIS disease changes was determined by single factor Logistic regression.Meanwhile,a Logistic regression prediction model based on ultra-high resolution CT signs was constructed,and receiver operating characteristic(ROC)curve of the prediction model was drawn.Results Single factor analysis,The proportion of＞60 years old patients in the outcome group,the level of serum CEA,CY211,NSE,tumor diameter,proportion of irregular shape,uneven density,non-vascular union,proportion of burrs,proportion of pleural traction,proportion of vacuoles,proportion of air bronchi,proportion of pleural indentation,proportion of multiple lesions,and proportion of smoking history in the ultra-high resolution CT signs were all lower than those in the adverse group.The proportion of regular shape,uniform density and vascular connection in CT signs was higher than that in bad group(P＜0.05).Logistic regression analysis showed that age＞60 years old(high proportion),tumor diameter(large),irregular shape,burr,uneven density,pleural traction,vasculature,vacuole sign and multiple lesions were risk factors affecting the prognosis of AIS patients(P＜0.05).Logistic regression prediction model was constructed based on ultra-high resolution CT signs(morphology,burr,density,pleural traction,vascular relationship,vacuole sign):Logistic(P)=-15.479+1.116X1+1.266X2+0.983X3+1.062X4+0.997X5+1.024X6,Among them,X1 was irregular shape,X2 was burr,X3 was uneven density,X4 was pleural traction,X5 was vascular connection,and X6 was vacuolar sign.The sensitivity,accuracy,positive predictive value and AUC of this model in predicting the prognosis of AIS patients were 93.65％,90.77％,90.26％,and 0.933,respectively.Conclusion The establishment of AIS clinical outcome prediction model based on ultra-high resolution CT signs can predict the clinical outcome more accurately.
Ultra-high Resolution CT SignsAdenocarcinoma of Lung in SituClinical OutcomePrediction ModelPredictive Value
万方数据
维普
