Study on predicting breast cancer HER-2 status through full-field digital mammography radiomic and deep learning features
HE Fei
HUANG Zhongjiang
WU Peizeng
GUO Xiaofen
WANG Lei
Abstract：Objective To predict the HER-2 status in breast cancer by integrating full-field digital mammography(FFDM)radiomics features with deep learning features.Methods Retrospective analysis was conducted on of breast cancer patients who underwent clinical surgery or biopsy from March 2018 to December 2023 at the Affiliated Hospital of Shanxi University of Chinese Medicine.Regions of interest within FFDM images were manually delineated to extract radiomics and deep learning features.Following LASSO-based feature selection,support vector machine algorithms were used to construct both a radiomics model and a deep learning model.A composite model was developed using multivariate logistic regression analysis.The predictive performance of each model was evaluated by calculating their AUC values,and their effectiveness and practical value in real clinical decision-making were assessed using decision curve analysis curves.Results The radiomics model exhibited AUC values of 0.835(95%CI:0.761-0.898)in the training set and 0.781(95%CI:0.701-0.857)in the test set.The deep learning model demonstrated AUC values of 0.904(95%CI:0.855-0.955)in the training set and 0.883(95%CI:0.823-0.939)in the test set.The composite model achieved AUC values of 0.921(95%CI:0.872-0.967)in the training set and 0.890(95%CI:0.828-0.942)in the test set.Decision curve analysis indicated that all three models provided a greater net benefit compared to assuming all cases as either HER-2 positive or negative,with the composite model offering the maximum net benefit at various risk thresholds.Conclusion The integration of FFDM radiomics and deep learning features significantly enhances the prediction of HER-2 status in breast cancer,greatly improving both accuracy and reliability.This advancement opens new avenues for the diagnosis and treatment of breast cancer.
Keywords：breast cancerradiomicsdeep learningHER-2 status
Publication Date：2024-08-28
Online Publishing Date：2025-08-15(First online date of this platform, not the publication date of the document)
Pages：7( 804-810 )