Deep Learning Model Based on Dual-Modal Ultrasound and Molecular Data for Predicting Response to Neoadjuvant Chemotherapy in Breast Cancer
Jia-Xin Huang,Jun Shi,Sai-Sai Ding,Hui-Li Zhang,Xue-Yan Wang,Shi-Yang Lin,Yan-Fen Xu,Ming-Jie Wei,Long-Zhong Liu,Xiao-Qing Pei
DOI: https://doi.org/10.1016/j.acra.2023.03.036
IF: 5.482
2023-06-03
Academic Radiology
Abstract:Rationale and Objectives To carry out radiomics analysis/deep convolutional neural network (CNN) based on B-mode ultrasound (BUS) and shear wave elastography (SWE) to predict response to neoadjuvant chemotherapy (NAC) in breast cancer patients. Materials and Methods In this prospective study, 255 breast cancer patients who received NAC between September 2016 and December 2021 were included. Radiomics models were designed using a support vector machine classifier based on US images obtained before treatment, including BUS and SWE. And CNN models also were developed using ResNet architecture. The final predictive model was developed by combining the dual-modal US and independently associated clinicopathologic characteristics. The predictive performances of the models were assessed with five-fold cross-validation. Results Pretreatment SWE performed better than BUS in predicting the response to NAC for breast cancer for both the CNN and radiomics models ( P < 0.001). The predictive results of the CNN models were significantly better than the radiomics models, with AUCs of 0.72 versus 0.69 for BUS and 0.80 versus 0.77 for SWE, respectively ( P = 0.003). The CNN model based on the dual-modal US and molecular data exhibited outstanding performance in predicting NAC response, with an accuracy of 83.60% ± 2.63%, a sensitivity of 87.76% ± 6.44%, and a specificity of 77.45% ± 4.38%. Conclusion The pretreatment CNN model based on the dual-modal US and molecular data achieved excellent performance for predicting the response to chemotherapy in breast cancer. Therefore, this model has the potential to serve as a non-invasive objective biomarker to predict NAC response and aid clinicians with individual treatments.
radiology, nuclear medicine & medical imaging
What problem does this paper attempt to address?