Intratumoral And Peritumoral Radiomics Assessment Of HER2 Status In Breast Cancer Based On DBT

Objective:To investigate the application value of intratumoral and peritumoral radiomics based on Digital Breast Tomosynthesis(DBT)for HER2 status in breast cancer,and the application value of a comprehensive model based on clinical-radiomics features to assist patients in clinical decision making.Methods:Imaging,clinical and pathological data of female breast cancer patients who received breast DBT examination before treatment and were confirmed by puncture biopsy or postoperative pathology from September 2020 to March 2022 were collected.The HER2 status of each patient was determined according to immunohistochemical staining(IHC)and in situ hybridization(ISH)results.The tumor region of interest(ROI)was delineated on the DBT images of the CC and MLO positions,and the intratumoral ROI was automatically expanded by 3 mm to obtain the peritumoral region,and the intratumoral and peritumoral three-dimensional volumetric region of interest(VOI)was generated.Extract and screen the radiomics features,and then establish three radiomics models of intratumoral,peritumoral,and combined intratumoral and peritumoral model.Statistical analysis of clinicopathological data was performed to build clinical models.Then,a comprehensive model based on clinical-radiomics was established and a nomogram was drawn.Statistical indicators such as ROC curve and area under curve(AUC)were used to evaluate the performance of each model,and calibration curve and decision curve were drawn to evaluate the clinical application value of the comprehensive model.Results:In this study,178 patients with breast cancer were eventually included,including 66HER2-positive patients and 112 negative patients.AUC of intratumoral,peritumoral,and combined intratumoral and peritumoral radiomics models were 0.800,0.726 and0.844,respectively.After single and multi-factor logistic regression analysis,suspected malignant calcification was identified as clinical feature,AUC of clinical model and combined clinical-imaging model were 0.585 and 0.850,respectively.Conclusion:DBT-based radiomics features can be used to assess HER2 status of breast cancer,and combined with peritumoral radiomics features can improve model performance.And the comprehensive model and nomogram combined with clinical and radiomics features have good predictive value.