Abstract #1432

Background Parenchymal Enhancement Radiomic Features for Neoadjuvant Treatment Response Prediction in Breast Cancer Patients

Alex Anh-Tu Nguyen¹, Natsuko Onishi¹, Wen Li¹, Deep K Hathi¹, Rohan Nadkarni¹, Efstathios D Gennatas², Ella F Jones¹, I-SPY2 Consortium³, David C Newitt¹, and Nola M Hylton¹

¹Radiology & Biomedical Imaging, University of California, San Francisco, San Francisco, CA, United States, ²Epidemiology & Biostatistics, University of California, San Francisco, San Francisco, CA, United States, ³Quantum Leap Healthcare Collaborative, San Francisco, CA, United States

Background parenchymal enhancement (BPE) observed in normal fibroglandular tissue in breast dynamic contrast-enhanced MRI shows an association with breast cancer risk and has been investigated to predict treatment response. Contralateral BPE radiomic features and mean BPE values were calculated from DCE-MRI and their predictive performance of pathologic complete response after neoadjuvant chemotherapy was tested. The best mean AUCs were found using BPE radiomics based random forest model with minimal-redundancy-maximal-relevance feature selection in the HR- group and HER2- group. Our results show that contralateral BPE radiomic features have potential as imaging biomarkers for treatment response.

This abstract and the presentation materials are available to members only; a login is required.

Join Here