Share：
Share this content in WeChat
X
[Chinese] [PDF] 663 3
• Clinical Article •
Development and validation of an intratumoral and peritumoral deep learning radiomics model based on DCE-MRI for predicting the response to neoadjuvant chemotherapy in triple-negative breast cancer: A multicenter study
ZHANG Aoqi  ZHU Qiuyang  LIU Mengling  LI Ran  ZHU Yun  TANG Xiaomin  ZHAO Cancan  HE Jie  XIE Zongyu 

Cite this article as: ZHANG A Q, ZHU Q Y, LIU M L, et al. Development and validation of an intratumoral and peritumoral deep learning radiomics model based on DCE-MRI for predicting the response to neoadjuvant chemotherapy in triple-negative breast cancer: A multicenter study[J]. Chin J Magn Reson Imaging, 2025, 16(9): 96-104. DOI:10.12015/issn.1674-8034.2025.09.015.


[Abstract] Objective To investigate the value of an intratumoral and peritumoral deep learning radiomics (DLR) model based on dynamic contrast-enhanced MRI (DCE-MRI) for predicting neoadjuvant chemotherapy (NAC) response in triple-negative breast cancer (TNBC).Materials and Methods This retrospective study enrolled 161 TNBC patients from two medical centers who underwent pre-NAC DCE-MRI. Data from center 1 (n = 112) served as the training set, while data from center 2 (n = 49) constituted an external validation cohort. An additional public TICA dataset (n = 74) was used as an independent external test set. The study comprised three components: traditional radiomics, deep learning (DL), and integrated DLR modeling. Radiomic features were extracted from both the intratumoral region and the peritumoral rims at 3 mm, 5 mm, and 7 mm distances using PyRadiomics. Performance of seven classifiers, support vector machine (SVM), k-nearest neighbors (KNN), extreme gradient boosting (XGBoost), extra trees (ET), logistic regression (LR), random forest (RF), and naive bayes (NB), was evaluated across different feature models to identify the optimal classifier for constructing the radiomics model. DL was implemented with a 3D DenseNet-121 backbone. Following the training and prediction of the DL model, the generated deep learning score (DL_score) was fused with the radiomics features. XGBoost, selected as the optimal classifier, was then used to build the final DLR fusion model. Diagnostic performance was assessed via receiver operating characteristic (ROC) curves and area under the curve (AUC). Calibration curves evaluated model fit, while decision curve analysis (DCA) quantified clinical utility.Results The optimal radiomics model was an ExtraTrees-based model combining intra-tumoral and peri-tumoral 3 mm region features, with AUC values of 0.847, 0.780, and 0.720 in the training, external validation, and external test sets, respectively. The DL model outperformed the radiomics model in identifying the response to NAC in TNBC patients, with AUC values of 0.865, 0.810, and 0.820 in the training, external validation, and external test sets, respectively. Compared to a single model, DLR further improved the discriminative ability, with AUC and accuracy of 0.917, 0.898, and 0.886, and 90.1%, 87.9%, and 86.5% in the training, external validation, and external test sets, respectively, demonstrating better clinical benefits and good calibration.Conclusions The DLR fusion model, integrating intratumoral and peritumoral deep-learning radiomic features derived from DCE-MRI, demonstrates potential clinical utility for predicting NAC response in TNBC patients.
[Keywords] triple-negative breast cancer；radiomics；deep learning；neoadjuvant therapy；magnetic resonance imaging

ZHANG Aoqi1, 2   ZHU Qiuyang3   LIU Mengling1, 2   LI Ran1, 2   ZHU Yun1   TANG Xiaomin1   ZHAO Cancan1   HE Jie4   XIE Zongyu1*  

1 Department of Radiology, the First Affiliated Hospital of Bengbu Medical University, Bengbu 233004, China

2 School of Medical Imaging, Bengbu Medical University, Bengbu 233030, China

3 Department of Cybersecurity, School of Information and Network Engineering, Anhui University of Science and Technology, Bengbu 233030, China

4 Department of Radiology, Sir Run Run Shaw Hospital, School of Medicine, Zhejiang University, Hangzhou 310016, China

Corresponding author: XIE Z Y, E-mail: zongyuxie@sina.com

Conflicts of interest   None.

Received  2025-06-19
Accepted  2025-08-28
DOI: 10.12015/issn.1674-8034.2025.09.015
Cite this article as: ZHANG A Q, ZHU Q Y, LIU M L, et al. Development and validation of an intratumoral and peritumoral deep learning radiomics model based on DCE-MRI for predicting the response to neoadjuvant chemotherapy in triple-negative breast cancer: A multicenter study[J]. Chin J Magn Reson Imaging, 2025, 16(9): 96-104. DOI:10.12015/issn.1674-8034.2025.09.015.

[1]
TONG L, YU X L, WANG S, et al. Research progress on molecular subtyping and modern treatment of triple-negative breast cancer[J/OL]. Breast Cancer (Dove Med Press), 2023, 15: 647-658 [2025-06-18]. https://pubmed.ncbi.nlm.nih.gov/37644916/. DOI: 10.2147/BCTT.S426121.
[2]
YIN L, DUAN J J, BIAN X W, et al. Triple-negative breast cancer molecular subtyping and treatment progress[J/OL]. Breast Cancer Res, 2020, 22(1): 61 [2025-06-18]. https://pubmed.ncbi.nlm.nih.gov/32517735/. DOI: 10.1186/s13058-020-01296-5.
[3]
DUAN S, DILIMULATI A, WANG H Y, et al. Influencing factors of pathological complete response in triple-negative breast cancer with neoadjuvant chemotherapy[J]. Chin J Clin Res, 2024, 37(3): 354-358. DOI: 10.13429/j.cnki.cjcr.2024.03.006.
[4]
BIANCHINI G, DE ANGELIS C, LICATA L, et al. Treatment landscape of triple-negative breast cancer: expanded options, evolving needs[J]. Nat Rev Clin Oncol, 2021, 19(2): 91-113. DOI: 10.1038/s41571-021-00565-2.
[5]
DENKERT C, VON MINCKWITZ G, DARB-ESFAHANI S, et al. Tumour-infiltrating lymphocytes and prognosis in different subtypes of breast cancer: a pooled analysis of 3771 patients treated with neoadjuvant therapy[J]. Lancet Oncol, 2018, 19(1): 40-50. DOI: 10.1016/S1470-2045(17)30904-X.
[6]
ZENG Q, LIU L, HE C W, et al. Early and noninvasive prediction of response to neoadjuvant therapy for breast cancer via longitudinal ultrasound and MR deep learning: a multicentre study[J]. Acad Radiol, 2025, 32(3): 1264-1273. DOI: 10.1016/j.acra.2024.10.033.
[7]
GRADISHAR W J, MORAN M S, ABRAHAM J, et al. NCCN guidelines® insights: breast cancer, version 4.2023[J]. J Natl Compr Canc Netw, 2023, 21(6): 594-608. DOI: 10.6004/jnccn.2023.0031.
[8]
ZHOU J Y, YOU C, WANG Z Z, et al. Value of radiomics model based on spatiotemporal heterogeneity of MRI to predict pathological complete response in triple-negative breast cancer[J]. Chin J Magn Reson Imag, 2024, 15(1): 28-34. DOI: 10.12015/issn.1674-8034.2024.01.005.
[9]
LEITHNER D, WENGERT G J, HELBICH T H, et al. Clinical role of breast MRI now and going forward[J]. Clin Radiol, 2018, 73(8): 700-714. DOI: 10.1016/j.crad.2017.10.021.
[10]
ZHAO N N, ZHU Y, TANG X M, et al. Prediction of axillary lymph node metastasis in breast cancer based on intra-tumoral and peri-tumoral MRI radiomics nomogram[J]. Chin J Magn Reson Imag, 2023, 14(3): 81-87, 94. DOI: 10.12015/issn.1674-8034.2023.03.014.
[11]
XIE T W, GONG J, ZHAO Q F, et al. Development and validation of peritumoral vascular and intratumoral radiomics to predict pathologic complete responses to neoadjuvant chemotherapy in patients with triple-negative breast cancer[J/OL]. BMC Med Imaging, 2024, 24(1): 136 [2025-06-18]. DOI: 10.12015/issn.1674-8034.2023.03.014..DOI:10.1186/s12880-024-01311-7.
[12]
TAN R, SUI C X, WANG C, et al. MRI-based intratumoral and peritumoral radiomics for preoperative prediction of glioma grade: a multicenter study[J/OL]. Front Oncol, 2024, 14: 1401977 [2025-06-18]. https://pubmed.ncbi.nlm.nih.gov/38803534/. DOI: 10.3389/fonc.2024.1401977.
[13]
PANTHI B, MOHAMED R M, ADRADA B E, et al. Longitudinal dynamic contrast-enhanced MRI radiomic models for early prediction of response to neoadjuvant systemic therapy in triple-negative breast cancer[J/OL]. Front Oncol, 2023, 13: 1264259 [2025-06-18]. https://pubmed.ncbi.nlm.nih.gov/37941561/. DOI: 10.3389/fonc.2023.1264259.
[14]
MOHAMED R M, PANTHI B, ADRADA B E, et al. Multiparametric MRI-based radiomic models for early prediction of response to neoadjuvant systemic therapy in triple-negative breast cancer[J/OL]. Sci Rep, 2024, 14(1): 16073 [2025-06-18]. https://pubmed.ncbi.nlm.nih.gov/38992094/. DOI: 10.1038/s41598-024-66220-9.
[15]
HUANG Y H, ZHU T, ZHANG X L, et al. Longitudinal MRI-based fusion novel model predicts pathological complete response in breast cancer treated with neoadjuvant chemotherapy: a multicenter, retrospective study[J/OL]. EClinicalMedicine, 2023, 58: 101899 [2025-06-18]. https://pubmed.ncbi.nlm.nih.gov/37007742/. DOI: 10.1016/j.eclinm.2023.101899.
[16]
GIULIANO A E, EDGE S B, HORTOBAGYI G N. Eighth edition of the AJCC cancer staging manual: breast cancer[J]. Ann Surg Oncol, 2018, 25(7): 1783-1785. DOI: 10.1245/s10434-018-6486-6.
[17]
COATES A S, WINER E P, GOLDHIRSCH A, et al. Tailoring therapies: improving the management of early breast cancer: st Gallen international expert consensus on the primary therapy of early breast cancer 2015[J]. Ann Oncol, 2015, 26(8): 1533-1546. DOI: 10.1093/annonc/mdv221.
[18]
CHATTERJI M, MERCADO C L, MOY L. Optimizing 1.5-Tesla and 3-Tesla dynamic contrast-enhanced magnetic resonance imaging of the breasts[J]. Magn Reson Imaging Clin N Am, 2010, 18(2): 207-224, viii. DOI: 10.1016/j.mric.2010.02.011.
[19]
LUO S W, CHEN X B, YAO M X, et al. Intratumoral and peritumoral ultrasound-based radiomics for preoperative prediction of HER2-low breast cancer: a multicenter retrospective study[J/OL]. Insights Imaging, 2025, 16(1): 53 [2025-06-18]. https://pubmed.ncbi.nlm.nih.gov/40053171/. DOI: 10.1186/s13244-025-01934-6.
[20]
ZHANG S H, WANG X L, YANG Z, et al. Intra- and peritumoral radiomics model based on early DCE-MRI for preoperative prediction of molecular subtypes in invasive ductal breast carcinoma: a multitask machine learning study[J/OL]. Front Oncol, 2022, 12: 905551 [2025-06-18]. https://pubmed.ncbi.nlm.nih.gov/35814460/. DOI: 10.3389/fonc.2022.905551.
[21]
ZHOU J, YU X, WU Q X, et al. Radiomics analysis of intratumoral and different peritumoral regions from multiparametric MRI for evaluating HER2 status of breast cancer: a comparative study[J/OL]. Heliyon, 2024, 10(7): e28722 [2025-06-18]. https://pubmed.ncbi.nlm.nih.gov/38623231/. DOI: 10.1016/j.heliyon.2024.e28722.
[22]
LI B, KANG Z Q, MA Y Y, et al. Habitat radiomics based on pre-treatment DCE-MRI images for predicting pathological complete response(pCR) in invasive breast cancer after neoadjuvant chemotherapy[J]. J Clin Radiol, 2025, 44(7): 1211-1218. DOI: 10.13437/j.cnki.jcr.2025.07.005.
[23]
ZHANG R, JIANG C F, LI F, et al. Preoperative MRI-based radiomics analysis of intra- and peritumoral regions for predicting CD3 expression in early cervical cancer[J/OL]. Sci Rep, 2025, 15: 26754 [2025-06-18]. https://pubmed.ncbi.nlm.nih.gov/40702236/. DOI: 10.1038/s41598-025-12162-9.
[24]
TU Y S, WENG A T, REN A L, et al. Multimodal MRI manifestations and pathological basis of peritumoral infiltration of glioma in rat[J]. Chin J Magn Reson Imag, 2023, 14(1): 105-110. DOI: 10.12015/issn.1674-8034.2023.01.019.
[25]
BOISSIÈRE-MICHOT F, CHATEAU M C, THÉZENAS S, et al. Prognostic value of tertiary lymphoid structures in triple-negative breast cancer: integrated analysis with the tumor microenvironment and clinicopathological features[J/OL]. Front Immunol, 2024, 15: 1507371 [2025-06-18]. https://pubmed.ncbi.nlm.nih.gov/39723208/. DOI: 10.3389/fimmu.2024.1507371.
[26]
NIU S X, JIANG W Y, ZHAO N N, et al. Intra- and peritumoral radiomics on assessment of breast cancer molecular subtypes based on mammography and MRI[J]. J Cancer Res Clin Oncol, 2022, 148(1): 97-106. DOI: 10.1007/s00432-021-03822-0.
[27]
LEE H J, LEE J H, LEE J E, et al. Prediction of early clinical response to neoadjuvant chemotherapy in Triple-negative breast cancer: Incorporating Radiomics through breast MRI[J/OL]. Sci Rep, 2024, 14(1): 21691 [2025-06-18]. https://pubmed.ncbi.nlm.nih.gov/39289507/. DOI: 10.1038/s41598-024-72581-y.
[28]
SHI Z W, HUANG X M, CHENG Z L, et al. MRI-based quantification of intratumoral heterogeneity for predicting treatment response to neoadjuvant chemotherapy in breast cancer[J/OL]. Radiology, 2023, 308(1): e222830 [2025-06-18]. https://pubmed.ncbi.nlm.nih.gov/37432083/. DOI: 10.1148/radiol.222830.
[29]
QU Y H, ZHU H T, CAO K, et al. Prediction of pathological complete response to neoadjuvant chemotherapy in breast cancer using a deep learning (DL) method[J]. Thorac Cancer, 2020, 11(3): 651-658. DOI: 10.1111/1759-7714.13309.
[30]
PENG Y S, CHENG Z L, GONG C, et al. Pretreatment DCE-MRI-based deep learning outperforms radiomics analysis in predicting pathologic complete response to neoadjuvant chemotherapy in breast cancer[J/OL]. Front Oncol, 2022, 12: 846775 [2025-06-18]. https://pubmed.ncbi.nlm.nih.gov/35359387/. DOI: 10.3389/fonc.2022.846775.
[31]
ZHOU Z J, ADRADA B E, CANDELARIA R P, et al. Predicting pathological complete response to neoadjuvant systemic therapy for triple-negative breast cancers using deep learning on multiparametric MRIs[J/OL]. Annu Int Conf IEEE Eng Med Biol Soc, 2023, 2023: 1-4 [2025-06-18]. https://pubmed.ncbi.nlm.nih.gov/38083160/. DOI: 10.1109/EMBC40787.2023.10340987.
[32]
ADRADA B E, MOSELEY T W, KAPOOR M M, et al. Triple-negative breast cancer: histopathologic features, genomics, and treatment[J/OL]. Radiographics, 2023, 43(10): e230034 [2025-06-18]. https://pubmed.ncbi.nlm.nih.gov/37792593/. DOI: 10.1148/rg.230034.
[33]
BATOOL Z, KAMAL M A, SHEN B R. Advancements in triple-negative breast cancer sub-typing, diagnosis and treatment with assistance of artificial intelligence: a focused review[J/OL]. J Cancer Res Clin Oncol, 2024, 150(8): 383 [2025-06-18]. https://pubmed.ncbi.nlm.nih.gov/39103624/. DOI: 10.1007/s00432-024-05903-2.
[34]
GU J H, TONG T, HE C, et al. Deep learning radiomics of ultrasonography can predict response to neoadjuvant chemotherapy in breast cancer at an early stage of treatment: a prospective study[J]. Eur Radiol, 2022, 32(3): 2099-2109. DOI: 10.1007/s00330-021-08293-y.
[35]
WU L Y, LI S H, WU C J, et al. Ultrasound-based deep learning radiomics nomogram for differentiating mass mastitis from invasive breast cancer[J/OL]. BMC Med Imaging, 2024, 24(1): 189 [2025-06-18]. https://pubmed.ncbi.nlm.nih.gov/39060962/. DOI: 10.1186/s12880-024-01353-x.
[36]
ZHOU J Y, BAI Y S, ZHANG Y, et al. A preoperative radiogenomic model based on quantitative heterogeneity for predicting outcomes in triple-negative breast cancer patients who underwent neoadjuvant chemotherapy[J/OL]. Cancer Imaging, 2024, 24(1): 98 [2025-06-18]. https://pubmed.ncbi.nlm.nih.gov/39080809/. DOI: 10.1186/s40644-024-00746-z.

PREV Predictive value of DCE-MRI and TIC combined ADC models for the expression status of hormone receptors in mass-type breast cancer
NEXT Efficacy study of diffusion spectrum imaging-based habitat imaging in differentiating heterogeneity between clear cell renal cell carcinoma and fat-poor angiomyolipoma
  


LINK


Tel & Fax: +8610-67113815    E-mail: editor@cjmri.cn