Related Experiment Video
Updated: Jun 30, 2026
Tracking the Mammary Architectural Features and Detecting Breast Cancer with Magnetic Resonance Diffusion Tensor Imaging
Published on: December 15, 2014
Delta Radiomics Based on MRI for Predicting Axillary Lymph Node Pathologic Complete Response After Neoadjuvant
Ning Mao1, Yuhan Bao2, Chuntong Dong3
1Department of Radiology and Nuclear Medicine, Xuanwu Hospital Capital Medical University, Beijing, P R China (N.M., J.L.); Department of Radiology, Yantai Yuhuangding Hospital, Qingdao University, Yantai, Shandong, P R China (N.M., H.M., H.X., F.L.); Big Data and Artificial Intelligence Laboratory, Yantai Yuhuangding Hospital, Qingdao University, Yantai, Shandong, P R China (N.M., H.M., H.X., F.L.); Shandong Provincial Key Medical and Health Laboratory of Intelligent Diagnosis and Treatment for Women's Diseases (Yantai Yuhuangding Hospital), Shandong, P R China (N.M., H.Z., H.M., Q.W., H.X., F.L.).
Purpose:
To develop and test a radiomics nomogram based on magnetic resonance imaging (MRI) and clinicopathological factors for predicting the axillary pathologic complete response (apCR) to neoadjuvant chemotherapy (NAC) in breast cancer patients with axillary lymph node (ALN) metastases.
Materials And Methods:
A total of 319 patients who underwent MRI examination and received NAC treatment were enrolled from two centers, and the presence of ALN metastasis was confirmed by biopsy pathology before NAC. The radiomics features were extracted from regions of interest of ALNs before (pre-radiomics) and after (post-radiomics) NAC. The difference of features before and after NAC, named delta radiomics, was calculated. The variance threshold, selectKbest and least absolute shrinkage and selection operator algorithm were used to select radiomics features. Radscore was calculated by a linear combination of selected features, weighted by their respective coefficients. The univariate and multivariate logistic regression was used to select the clinicopathological factors and radscores, and a radiomics nomogram was built by multivariable logistic regression analysis. The performance of the nomogram was evaluated by the area under the receiver operator characteristic curve (AUC), decision curve analysis (DCA) and calibration curves. Furthermore, to explore the biological basis of radiomics nomogram, 16 patients with RNA-sequence data were included for genetic analysis.
Results:
The radiomics nomogram was constructed by two radscores (post- and delta- radscores) and one clinicopathological factor (progesterone hormone, PR), and showed powerful predictive performance in both internal and external test sets, with AUCs of 0.894 (95% confidence interval [CI], 0.877-0.959) and 0.903 (95% CI, 0.801-0.986), respectively. The calibration curves and DCA showed favorable consistency and clinical utility. With the assistance of nomogram, the rate of unnecessary ALND would be reduced from 60.42% to 21.88%, and the rate of final benefit rate would be increased from 39.58% to 70.83%. Moreover, genetic analysis revealed that high apCR prediction scores were associated with the upregulation of immune-mediated genes and pathways.
Conclusion:
The radiomics nomogram showed great performance in predicting apCR after NAC for breast cancer patients, which could help clinicians to identify patients with apCR and avoid unnecessary axillary lymph node dissection.
