Intra- and peri-tumoral radiomics based on dynamic contrast-enhanced MRI for prediction of benign disease in BI-RADS 4 breast lesions: A multicentre study.

e13542 Background: The study aimed to develop and optimize a radiomics model with features extracted from breast intra- and peri-tumoral region in dynamic contrast-enhanced (DCE) MRI to distinguish benign from malignant breast lesions of Breast Imaging Reporting and Data System (BI-RADS) 4, thereby obviating unnecessary biopsies. Methods: We retrospectively analyzed data of women with BI-RADS 4 or 5 lesions from three hospitals. The regions of interest (ROIs) of intra-tumoral regions were delineated manually on each slice of image. The peri-tumoral regions were then obtained by equidistant 3-dimensinal dilation of the tumor border with 1, 2 and 3mm, respectively. Seven classification models were built with features extracted from the intra- and peri-tumoral regions, including the Intra, Peri1mm, Peri2mm, Peri3mm, Intra+peri1mm, Intra+peri2mm and Intra+peri3mm. Diagnostic performance was evaluated by the area under the receiver operating characteristic curve (AUC) and compared by DeLong test. Subgroup analysis was performed after stratifying all lesions by enhancement pattern. To assess the rate of avoidable biopsies, three exploratory cutoff values on the ROC curve were examined. Results: The Comb2 model, built with features from peri-tumoral 2mm and intra-tumoral region, demonstrated the best performance with AUCs of 0·901 and 0·863 in the development and external test cohort, respectively. The Comb2 model was robust in both mass and non-mass enhancement lesions. At the three exploratory cutoff values on the ROC curve, the model identified 11·2% (sensitivity of C1≥90%), 49·5% (sensitivity of C2≥95%) and 63·3% (sensitivity of C3≥98%) of the benign lesions in the development cohort. Applying the identified cutoff values in the external test cohort showed the potential to lower the number of unnecessary biopsies of benign lesions. Conclusions: An MRI-based radiomics model built with features extracted from the intra-tumoral region combined with the peri-tumoral 2mm showed the best potential to reduce false-positive diagnoses and may avoid unnecessary biopsies with a low underestimate risk.