多模态影像联合临床因素列线图模型预测乳腺浸润性导管癌伴导管原位癌成分的价值OA

Objective:To investigate the predictive value of a nomogram model based on mammography(MG),color Dop-pler ultrasonography(CDUS)features and clinical factors for invasive ductal carcinoma coexisting with a ductal carcinoma in situ component(IDC-DCIS).Methods:Clinical and imaging data of 256 patients with surgically and pathologically confirmed invasive ductal carcinoma(IDC)were retrospectively analyzed.Patients were randomly allocated in a 7∶3 ratio into a training set(n=179,including 68 IDC-DCIS and 111 IDC)and a test set(n=77,including 31 IDC-DCIS and 46 IDC).Predictors of IDC-DCIS were screened through LASSO regression and multivariable Logistic regression analyses,culminating in the con-struction of nomogram models.Then,the predictive value of the model for IDC-DCIS was evaluated.Results:Patient age(OR=1.13,P＜0.001),premenopausal status(OR=25.94,P＜0.001),mammography-reported dense breasts(OR=3.03,P=0.010),mass with equal density(OR=4.99,P＜0.001),mass with suspicious calcifications(OR=3.39,P=0.002),and CDUS-reported hyper-echoic halo at the mass boundary(OR=3.59,P=0.003)were identified as significant predictors of IDC-DCIS.The nomogram model constructed demonstrated an area under the curve(AUC),sensitivity,and specificity of 0.828(95%CI:0.769～0.888),0.515,and 0.946 in the training set,respectively,and 0.823(95%CI:0.723～0.924),0.516,and 0.891 in the test set.Calibra-tion curves and decision curve analysis exhibited excellent concordance between the predicted and actual results,substantiat-ing the model's strong clinical applicability.Conclusion:The nomogram model based on MG,CDUS features,and clinical factors can intuitively and effectively predict the presence of a DCIS component in IDC patients preoperatively.