Diagnostic performance of intra-tumoral and peri-tumoral radiomics based on digital breast tomosynthesis imaging for the differentiation of benign and malignant breast lesions
Objective:To evaluate the differential diagnostic performance of intra-tumoral and peri-tumoral ra-diomics based on digital breast tomosynthesis(DBT)for breast lesions.Methods:The images of 135 female pa-tients who underwent DBT examination from July 2021 to December 2023 in the Second Affiliated Hospital of Nan-jing Medical University were retrospectively analyzed.The largest dimension of the tumor in the DBT image was se-lected to manually delineate the intra-tumoral region of interest(ROI),and the peri-tumoral ROI was obtained by expanding outward by 3 mm.Radiomics features were extracted and screened by Phython(Version 3.5).Support vector machine was used to construct intra-tumoral,peri-tumoral and intra-tumoral+peri-tumoral models.The dif-ferential diagnostic performance of the radiomics models was analyzed using the receiver operating characteristic(ROC)curve,calibration and decision curves.Results:A total of 135 female patients were analyzed including 55 benign and 80 malignant cases.10,5,and 14 optimal radiomics features were extracted and screened from intra-tumoral,peri-tumoral and intra-tumoral+peri-tumoral ROI to construct radiomics models.The AUC of intra-tumor-al,peri-tumoral and intra-tumoral+peri-tumoral radiomics model was 0.832,0.806,0.848 in the training set and 0.671,0.652,0.700 in the validation set.The calibration curve showed that the intra-tumoral+peri-tumoral ra-diomics model had good consistency with the ideal model in distinguishing benign and malignant breast lesions,and had good calibration ability.The decision curve analysis results also confirmed that it's clinical net benefit rate of was higher than others.Conclusion:DBT radiomics models added the peri-tumoural radiomics characteristics can be regarded as a effective approach to differentiate benign and malignant breast lesion for increased performance.
breast lesionsdigital breast tomosynthesisradiomicsintra-tumoralperi-tumoraldifferential di-agnosis
万方数据
维普
