Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Open Access

Peer-reviewed

Research Article

Diagnostic performance of DCE-MRI radiomics in predicting axillary lymph node metastasis in breast cancer patients: A meta-analysis

  • Fei Dong,

    Roles Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Writing – original draft, Writing – review & editing

    Affiliation Department of Medical Imaging, Yuncheng Central Hospital Affiliated to Shanxi Medical University, Yuncheng, Shanxi Province, China

    ⨯
  • Jie Li,

    Roles Data curation, Investigation, Methodology, Writing – original draft, Writing – review & editing

    Affiliation Department of Anesthesiology, Yuncheng Central Hospital Affiliated to Shanxi Medical University, Yuncheng, Shanxi Province, China

    ⨯
  • Junbo Wang,

    Roles Data curation, Investigation, Methodology, Writing – original draft, Writing – review & editing

    Affiliation Department of Medical Imaging, Yuncheng Central Hospital Affiliated to Shanxi Medical University, Yuncheng, Shanxi Province, China

    ⨯
  • Xiaohui Yang

    Roles Conceptualization, Methodology, Supervision, Writing – original draft, Writing – review & editing

    yangxiaohui19782@126.com

    Affiliation Department of Medical Imaging, Yuncheng Central Hospital Affiliated to Shanxi Medical University, Yuncheng, Shanxi Province, China

    ⨯

Abstract

Radiomics offers a novel strategy for the differential diagnosis, prognosis evaluation, and prediction of treatment responses in breast cancer. Studies have explored radiomic signatures from dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) for predicting axillary lymph node metastasis (ALNM) and sentinel lymph node metastasis (SLNM), but the diagnostic accuracy varies widely. To evaluate this performance, we conducted a meta-analysis performing a comprehensive literature search across databases including PubMed, EMBASE, SCOPUS, Web of Science (WOS), Cochrane Library, China National Knowledge Infrastructure (CNKI), Wanfang Data, and the Chinese BioMedical Literature Database (CBM) until March 31, 2024. The pooled sensitivity, specificity, positive likelihood ratio (PLR), negative likelihood ratio (NLR), diagnostic odds ratio (DOR), and the area under the receiver operating characteristic curve (AUC) were calculated. Twenty-four eligible studies encompassing 5588 breast cancer patients were included in the meta-analysis. The meta-analysis yielded a pooled sensitivity of 0.81 (95% confidence interval [CI]: 0.77–0.84), specificity of 0.85 (95%CI: 0.81–0.87), PLR of 5.24 (95%CI: 4.32–6.34), NLR of 0.23 (95%CI: 0.19–0.27), DOR of 23.16 (95%CI: 17.20–31.19), and AUC of 0.90 (95%CI: 0.87–0.92), indicating good diagnostic performance. Significant heterogeneity was observed in analyses of sensitivity (I2 = 74.64%) and specificity (I2 = 83.18%). Spearman’s correlation coefficient suggested no significant threshold effect (P = 0.538). Meta-regression and subgroup analyses identified several potential heterogeneity sources, including data source, integration of clinical factors and peritumor features, MRI equipment, magnetic field strength, lesion segmentation, and modeling methods. In conclusion, DCE-MRI radiomic models exhibit good diagnostic performance in predicting ALNM and SLNM in breast cancer. This non-invasive and effective tool holds potential for the preoperative diagnosis of lymph node metastasis in breast cancer patients.

Citation: Dong F, Li J, Wang J, Yang X (2024) Diagnostic performance of DCE-MRI radiomics in predicting axillary lymph node metastasis in breast cancer patients: A meta-analysis. PLoS ONE 19(12): e0314653. https://doi.org/10.1371/journal.pone.0314653

Editor: Peyman Tabnak, Tabriz University of Medical Sciences, ISLAMIC REPUBLIC OF IRAN

Received: July 8, 2024; Accepted: November 13, 2024; Published: December 3, 2024

Copyright: © 2024 Dong et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: All relevant data are within the article and its Supporting Information files.

Funding: The author(s) received no specific funding for this work.

Competing interests: The authors have declared that no competing interests exist.

Introduction

Breast cancer is one of the most common malignant tumors and a leading cause of cancer-related deaths mortality among women worldwide [1]. The status of axillary lymph node (ALN) is crucial for tumor staging, treatment decisions, and prognosis [2, 3]. Patients with axillary lymph node metastasis (ALNM) face a higher risk of distant metastasis and lower overall survival compared to those without node metastasis [4]. Standard methods for confirming ALN status, such as sentinel lymph node biopsy (SLNB) and axillary lymph node dissection (ALND), are invasive and carry risks like shoulder dysfunction, lymphedema, subcutaneous effusion, and nerve injury [5, 6]. Additionally the significant false-negative rate of SLNB complicates [7], highlighting the need for non-invasive and accurate diagnostic methods.

Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) offers detailed morphological and hemodynamic information about tumor lesions, making it a valuable tool in breast cancer staging [8]. Despite its potential pre-operative ALN assessment [9, 10], DCE-MRI’s diagnostic performance for ALNM is suboptimal. Evaluation based on qualitative or semi-qualitative features can be influenced by the radiologist’s expertise and experience [11]. Radiomics, which extracts high-throughput quantitative features from medical images within the region of interest (ROI), offers a novel and objective approach for comprehensive and detailed tumor heterogeneity assessment [12]. Radiomics has shown great potential in diagnosis, prognosis evaluation, and predicting therapeutic response by analyzing the relationships between various radiomic features and clinical information [13, 14].

Recently, interest has surged in applying DCE-MRI radiomics for assessing ALN status in breast cancer, with numerous models developed and validated for predicting ALNM and sentinel lymph node metastasis (SLNM) [9, 15–17]. However, inconsistencies across studies arise from variations in sample size, MRI protocols, radiomic features, and model construction. To address these disparities, we conducted a meta-analysis to systematically evaluate the diagnostic performance of DCE-MRI radiomics in predicting ALNM and SLNM in breast cancer.

Materials and methods

Literature search and selection

This meta-analysis adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guideline (S1 Checklist) [18]. Comprehensive searches were conducted in databases including PubMed, EMBASE, SCOPUS, Web of Science (WOS), Cochrane Library, China National Knowledge Infrastructure (CNKI), Wanfang Data, and the Chinese BioMedical Literature Database (CBM) for relevant studies published from up to March 31, 2024. The search utilized the following MeSH terms and variants: (‘Lymphatic Metastasis’[Mesh] OR ‘Lymph Nodes’[Mesh] OR ‘lymph’[Title/Abstract]) AND (‘Breast Neoplasms’[Mesh] OR ‘breast cancer’[Title/Abstract] OR ‘breast carcinoma’[Title/Abstract]) AND (‘Magnetic Resonance Imaging’[Mesh] OR ‘magnetic resonance imaging’[Title/Abstract] OR ‘MRI’[Title/Abstract]) AND (‘Radiomics’[Mesh] OR ‘radiomic*’[Title/Abstract]). Searches were restricted to English and Chinese articles. Additionally, references of relevant articles were manually reviewed to identify further eligible studies.

The inclusion criteria included: (1) studies involving breast cancer patients with definitive pathological outcomes for ALNM or SLNM; (2) studies establishing models to classify ALNM or SLNM based on DCE-MRI radiomics; (3) studies providing sufficient data for constructing a 2 × 2 contingency table to calculate sensitivity and specificity. Exclusions criteria included: (1) overlapping data; (2) incomplete or missing analytical data for constructing the 2 × 2 contingency table; (3) radiomics of MRI sequences other than DCE-MRI; (4) studies focusing on predicting ALN burden. Case reports, reviews, conference abstracts, and animal studies were also excluded. Two independent authors (JL, JW) performed the literature search and selection, and resolved any disagreements through discussion.

Data extraction

Two authors (FD, JL) independently extracted the following data from each study: first author, publication year, country/region, study design, sample size, data source, reference standard, MRI equipment, magnetic field strength, DCE phase, ROI, clinical factors, peritumoral features, and feature selection and model construction details. True positive (TP), false positive (FP), false negative (FN), and true negative (TN) values were also extracted to construct the 2 × 2 contingency table. Any conflicts were resolved through discussion.

Methodological quality assessment

Methodological quality was assessed using the Quality Assessment of Diagnostic Accuracy Studies (QUADAS-2) tool [19]. Bias in patient selection, index test, reference standard, and flow and timing was rated as high, unclear, or low risk. Applicability concerns in patient selection, index test, and reference standard were rated as high, unclear, or low. In addition, the METhodological RadiomICs Score (METRICS), a scoring tool for quality assessment of radiomics research, was used to assess the methodological quality of included studies [20]. The quality was categorized according to METRICS score as very low (0 ≤ score < 20%), low (20 ≤ score < 40%), moderate (40 ≤ score < 60%), good (60 ≤ score < 80%), and excellent (80 ≤ score ≤ 100%). Two authors (FD, JL) independently assessed the methodological quality, and resolved conflicts through discussion.

Statistical analysis

Predictive accuracy of DCE-MRI radiomic model was assessed by calculating pooled sensitivity, specificity, positive likelihood ratio (PLR), negative likelihood ratio (NLR), and diagnostic odds ratio (DOR) with corresponding 95% confidence interval (95%CI). The summary receiver operating characteristic (SROC) curve was plotted. The area under the curve (AUC) was estimated to summarize diagnostic performance, with discrimination ability categorized as excellent (AUC>0.90), good (0.80–0.90), fair (0.70–0.80), poor (0.60–0.70), and unqualified (0.50–0.60). Heterogeneity among studies was assessed using the I2 statistic with Cochrane Q test, with I2 >50% and P<0.05 indicating significant heterogeneity, warranting a random-effect model for pooled analysis. In cases of significant heterogeneity, Spearman’s correlation coefficient was calculated to assess the threshold effect. Meta-regression and subgroup analyses were conducted to explore heterogeneity sources, considering covariates such as region, data source, outcome, combining clinical factors, combining peritumoral features, MRI equipment, magnetic field strength, ROI, lesion segmentation method, and classifier type. We used Leave-One-Out sensitivity analysis to test result robustness and Deek’s funnel plot to assess publication bias. Clinical utility of DCE-MRI radiomics was evaluated using the Fagan plot, calculating posttest probability of lymph node metastasis based on pretest probability. All analyses were conducted using STATA 16.0 (SataCorp, TX, USA). P value <0.05 was considered statistically significant.

Results

Baseline characteristics of studies included in meta-analysis

The literature search initially yielded 724 records, of which 407 unique articles remained after duplicate removal (Fig 1) Title and abstract screening led to the exclusion of 367 items. A total of 40 articles were selected for full-text review, with 16 ultimately excluded for various reasons. All the identified articles with eligibility status are listed in S1 Table. In total, 24 eligible studies comprising 5588 breast cancer patients were included in the meta-analysis [9, 10, 15–17, 21–39].

thumbnail
Download:
Fig 1. Flow diagram of literature search and selection.

https://doi.org/10.1371/journal.pone.0314653.g001

Among these studies, four were published in Chinese [36–39], while the remainder were in English. Except for two studies conducted outside China [15, 17], all were retrospective studies performed in China. Most studies were single-center; however, three recruited patients from two institutions [10, 24, 35]. Six studies developed models for SLNM prediction, while the others focused on ALNM. Ten studies incorporated combined clinical factors into radiomic-based models, and seven combined intratumoral and peritumoral features. The ROI was the breast tumor in 21 studies, the lymph node in two [24, 31], and both the tumor and lymph node in one [35]. The DCE phases analyzed included the strongest enhanced phase, first postcontrast phase (CE1), second postcontrast phase (CE2), and third postcontrast phase (CE3) in 7, 3, 9, and 5 studies, respectively. Nineteen studies manually segmented the ROI, while five studies used semi-automatic or automatic segmentation [15, 17, 20, 31, 32]. Logistic regression (LR) and support vector machine (SVM) were the most commonly used classifiers, featured in 11 and 9 studies, respectively. The baseline characteristics of all studies are summarized in Table 1, while details of MRI scanning and model construction are presented listed in Table 2. The extracted analytical data and corresponding 2 × 2 contingency tables showing TP, FP, FN, and TN for each study are presented in S2 Table.

thumbnail
Download:
Table 1. Baseline characteristics of studies included in the meta-analysis.

https://doi.org/10.1371/journal.pone.0314653.t001

thumbnail
Download:
Table 2. Features of MRI scanning and prediction model construction.

https://doi.org/10.1371/journal.pone.0314653.t002

Methodological quality assessment

In the QUADAS-2 assessment, the majority of studies exhibited unclear risk of bias in the patient selection domain due to insufficient descriptions of the patient selection process. One study showed high risk of bias in this domain [20], while four studies demonstrated low risk of bias [16, 26, 27, 29]. Regarding the index test, eight studies were considered to have unclear risk of bias due to the unknown use of a blinded setting [16, 20, 22, 26, 32, 33, 36, 39]. The remaining studies were rated with low risk of bias. Details of the risk of bias and applicability concerns are illustrated in Fig 2.

thumbnail
Download:
Fig 2. Methodological quality assessment of the included studies using the Quality Assessment of Diagnostic Accuracy Studies (QUADAS-2) tool.

https://doi.org/10.1371/journal.pone.0314653.g002

The METRICS assessment of included studies is detailed in S3 Table, with a median score of 69.0% (range: 53.1–90.3%). Three studies were categorized as moderate quality [17, 21, 23], 2 as excellent quality [10, 35], and the others as good quality.

Pooled analysis

As illustrated in Fig 3, the meta-analysis of 24 studies revealed that the pooled sensitivity and specificity for DCE-MRI radiomic signatures for predicting ALNM and SLNM in breast cancer patients were 0.81 (95%CI: 0.77–0.84) and 0.85 (95%CI: 0.81–0.87), respectively. The pooled analysis produced a PLR of 5.24 (95%CI: 4.32–6.34), an NLR of 0.23 (95%CI: 0.19–0.27), and a DOR of 23.16 (95%CI: 17.20–31.19). The SROC curve was plotted in Fig 4, yielding an AUC of 0.90 (95%CI: 0.87–0.92). Therefore, DCE-MRI radiomic signatures exhibited a good predictive ability for assessing the risk of ALNM and SLNM in breast cancer patients.

thumbnail
Download:
Fig 3. Forest plots of sensitivity and specificity of DCE-MRI radiomics in predicting lymph node metastasis in breast cancer.

https://doi.org/10.1371/journal.pone.0314653.g003

thumbnail
Download:
Fig 4. The summary receiver operating characteristics curve of the diagnostic performance of DCE-MRI radiomics in predicting lymph node metastasis in breast cancer.

https://doi.org/10.1371/journal.pone.0314653.g004

Heterogeneity

Significant heterogeneity was observed in the pooled analysis of sensitivity (I2 = 74.64%) and specificity (I2 = 83.18%).The Spearman’s correlation coefficient indicated no evidence of a threshold effect (coefficient = -0.131, P = 0.538). Univariate meta-regression analysis and subgroup analyses identified various factors contributing to heterogeneity, including data source, outcome assessment, clinical factors, peritumor features, MRI equipment, magnetic field, lesion segmentation, and classifier type (Table 3).

thumbnail
Download:
Table 3. Results of univariate meta-regression and subgroup analyses.

https://doi.org/10.1371/journal.pone.0314653.t003

Higher pooled sensitivity and specificity were observed in studies using single-center data compared to those using two-center data (sensitivity: 0.81 vs 0.78; specificity: 0.85 vs 0.81). Studies focused on ALNM exhibited greater sensitivity and specificity than those on SLNM (sensitivity: 0.82 vs 0.78; specificity: 0.85 vs 0.84), Additionally, studies adopting SVM for prediction models showed improved sensitivity and specificity compared to those using LR (sensitivity: 0.87 vs 0.77; specificity: 0.86 vs 0.85). Models than incorporated clinical factors displayed higher specificity (0.86 vs 0.83) but lower sensitivity (0.79 vs 0.82) compared to radiomic models alone. Furthermore, models integrating both intratumoral and peritumor features demonstrated higher sensitivity (0.84 vs 0.79) and but slightly lower specificity (0.83 vs 0.85) than those using only intratumoral features. 3.0T imaging had higher sensitivity (0.83 vs 0.74) but slightly reduced specificity (0.84 vs 0.86) compared to 1.5T imaging. Siemens MRI equipment was associated with higher sensitivity (0.87 vs 0.76) but lower specificity (0.80 vs 0.87) than GE equipment. Semi-automatic or automatic segmentation improved sensitivity (0.86 vs 0.80) but lowered the specificity (0.80 vs 0.85) compared to manual segmentation. No significant differences in pooled sensitivity and specificity were observed between studies using only breast tumors as the ROI and those using lymph nodes (P>0.05).

Sensitivity analysis and publication bias

A sensitivity analysis was conducted to assess the influence of each study on overall results (S4 Table). The pooled estimates for sensitivity, specificity, PLR, NLR, DOR, and AUC of the SROC curve remained stable after the exclusion of any single study, confirming the robustness of the findings. Publication bias was assessed using a funnel plot and asymmetry test, which showed no significant publication bias (P = 0.267, Fig 5).

thumbnail
Download:
Fig 5. Deek’s funnel plot asymmetry test for publication bias.

https://doi.org/10.1371/journal.pone.0314653.g005

Clinical utility

Using a DCE-MRI radiomics model, the posttest probability would rise from 20% to 57% with a PLR of 5 when the pretest result was positive (Fig 6). Conversely, the posttest probability would drop from 20% to 5% with an NLR of 0.23 when the pretest result was negative (Fig 6).

thumbnail
Download:
Fig 6. Fagan plot for assessing the clinical utility.

https://doi.org/10.1371/journal.pone.0314653.g006

Discussion

In this meta-analysis, we incorporated data from 24 studies encompassing 5588 breast cancer patients to assess the diagnostic performance of DCE-MRI radiomics in characterizing ALNM and SLNM. The prediction accuracy of DCE-MRI radiomics was notable, with pooled sensitivity, specificity, DOR, and overall AUC of 0.81 (95%CI: 0.77–0.84), 0.85 (95%CI: 0.81–0.87), 23.16 (95%CI: 17.20–31.19), and 0.90 (95%CI: 0.87–0.92), respectively. These findings indicate that DCE-MRI radiomics is an effective and accurate tool for predicting lymph node metastasis, holding significant clinical value for personalized treatment strategies in breast cancer patients.

However, substantial heterogeneity was noted in the sensitivity (I2 = 74.64%) and specificity (I2 = 83.18%) analyses. The threshold effect was ruled out as a source of heterogeneity, as indicated by the non-significant Spearman’s correlation coefficient (P = 0.538). Univariate meta-regression and subgroup analyses identified potential sources of heterogeneity. Notably, most studies were based on single-center data, with only three performing external validation or testing [10, 24, 35]. Diagnostic performance was lower in two-center studies compared to single-center studies, with reduced sensitivity and specificity. These findings indicated that the reproducibility of radiomic models developed by these single-center studies requires validation across multiple external datasets.

DCE-MRI radiomic models demonstrated lower pooled sensitivity and specificity for SLNM prediction compared to ALNM prediction. The future development of more effective tolls for SLNM prediction is essential. Most studies utilized SVM or LR algorithms to construct prediction models. SVM, which is robust to model misspecification and handles high-dimensional data effectively [40], showed better predictive ability than LR in subgroup analysis, with higher sensitivity and slightly higher specificity. Thus, SVM appears preferable over LR for radiomic model construction.

Peritumoral regions provide valuable insight into the tumor microenvironment relevant to tumor growth and invasion [41, 42]. Peritumoral radiomics have been effective in predicting breast lesion malignancy [43], molecular subtype [44], and response to neoadjuvant chemotherapy [45]. Most radiomic studies have focused on only intratumoral regions without considering peritumoral radiomic signatures. This meta-analysis included seven studies that combined intratumoral and peritumoral radiomic signatures for predicting tumor metastasis [16, 22, 26, 30, 34, 37, 38]. Among these studies, combined models demonstrated higher diagnostic yield than those using only intratumoral or peritumoral signatures [16, 30, 34, 37]. Subgroup analysis of our study showed that the combined radiomic models had higher sensitivity (0.84 vs 0.79) and only slightly lower specificity (0.83 vs 0.85) compared to intratumoral radiomic models. The pooled DOR for combined models was higher than that for intratumoral models (25.92 [95%CI: 13.44–50.00] vs 22.12 [95%CI: 16.03–30.53]), suggesting that adding peritumoral signatures enhances diagnostic performance of DCE-MRI radiomics.

Imaging at 3.0T offers a higher signal-to-noise ratio, resulting in better spatial resolution and image quality compared to 1.5T [46]. This meta-analysis found that 3.0T imaging had significantly higher sensitivity (0.83 vs 0.74) and slightly lower specificity (0.84 vs 0.86) than 1.5T. The DOR value was also higher for 3.0T imaging than that for 1.5T imaging (26.74 [95%CI: 18.11–39.49] vs 16.59 [95%CI: 12.46–22.09]), indicating superior discriminating ability of 3.0T imaging.

Our meta-analysis has several advantages over previous studies. Earlier meta-analyses focused on multiple MRI sequences, including T1-weighted image (T1WI), T2-weighted fat-suppressed (T2-FS), diffusion-weighted imaging (DWI), and DCE-MRI [47–50]. Diagnostic performance can vary significantly across different MRI sequences. For example, Chen C et al. found that the DCE sequence had higher pooled sensitivity and DOR than T2-FS and DWI [47]. A recent meta-analysis focused on DCE-MRI radiomics but included only 13 studies and 1618 participants [51]. Our study, focusing solely on DCE-MRI, provides more specific and valuable clinical guidance. Additionally, our study, with a 3.5-fold larger sample size, demonstrated better diagnostic yields and narrower confidence intervals compared to previous meta-analysis [51], leading to more reliable conclusions regarding the diagnostic performance of DCE-MRI radiomics in predicting ALNM/SLNM.

Nonetheless, this meta-analysis has limitations. All studies are retrospective, which introduce potential selection bias. Most studies were single-center and conducted in Chinese populations, lacking external validation in multiple centers and in other populations. Therefore, the reproducibility and generalizability of radiomic models need future investigations in prospective, multi-center studies across different populations. Variations in image processing aspects, such as ROIs, MRI protocols, DCE phases, tumor segmentation, feature selection, and model construction, contribute to substantial heterogeneity and reduce reproducibility. Standards and best practice need to be established. Additionally, we selected the best-performing model from each study, which may overestimate diagnostic performance.

Conclusions

Our meta-analysis indicates that DCE-MRI radiomic models have good diagnostic performance in predicting ALNM and SLNM in breast cancer patients. Future prospective, large-scale, multi-center studies are needed to validate the effectiveness and clinical utility of this non-invasive method.

Supporting information

S1 Checklist. PRISMA-DTA checklist.

https://doi.org/10.1371/journal.pone.0314653.s001

(DOCX)

S1 Table. List of articles identified by literature search.

https://doi.org/10.1371/journal.pone.0314653.s002

(DOCX)

S2 Table. Analytical data extracted from included studies.

https://doi.org/10.1371/journal.pone.0314653.s003

(XLSX)

S3 Table. Quality assessment using METRICS.

https://doi.org/10.1371/journal.pone.0314653.s004

(XLSX)

S4 Table. Results of sensitivity analysis.

https://doi.org/10.1371/journal.pone.0314653.s005

(DOCX)

Acknowledgments

We thank all the participants of all the studies.

References

  1. 1. Sung H, Ferlay J, Siegel RL, Laversanne M, Soerjomataram I, Jemal A, et al. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA Cancer J Clin. 2021;71(3):209–49. pmid:33538338
  2. 2. Andersson Y, Bergkvist L, Frisell J, de Boniface J. Long-term breast cancer survival in relation to the metastatic tumor burden in axillary lymph nodes. Breast Cancer Res Treat. 2018;171(2):359–69. pmid:29846847
  3. 3. Tamirisa N, Thomas SM, Fayanju OM, Greenup RA, Rosenberger LH, Hyslop T, et al. Axillary Nodal Evaluation in Elderly Breast Cancer Patients: Potential Effects on Treatment Decisions and Survival. Ann Surg Oncol. 2018;25(10):2890–2898. pmid:29968029
  4. 4. Tonellotto F, Bergmann A, de Souza Abrahao K, de Aguiar SS, Bello MA, Thuler LCS. Impact of Number of Positive Lymph Nodes and Lymph Node Ratio on Survival of Women with Node-Positive Breast Cancer. Eur J Breast Health. 2019;15(2):76–84. pmid:31001608
  5. 5. Abass MO, Gismalla MDA, Alsheikh AA, Elhassan MMA. Axillary Lymph Node Dissection for Breast Cancer: Efficacy and Complication in Developing Countries. J Glob Oncol. 2018;4:1–8. pmid:30281378
  6. 6. Lyman GH, Somerfield MR, Bosserman LD, Perkins CL, Weaver DL, Giuliano AE. Sentinel Lymph Node Biopsy for Patients With Early-Stage Breast Cancer: American Society of Clinical Oncology Clinical Practice Guideline Update. J Clin Oncol. 2017;35(5):561–4. pmid:27937089
  7. 7. Harrison B. Update on sentinel node pathology in breast cancer. Semin Diagn Pathol. 2022;39(5):355–366. pmid:35803776
  8. 8. Baltzer PAT, Bickel H, Spick C, Wengert G, Woitek R, Kapetas P, et al. Potential of Noncontrast Magnetic Resonance Imaging With Diffusion-Weighted Imaging in Characterization of Breast Lesions: Intraindividual Comparison With Dynamic Contrast-Enhanced Magnetic Resonance Imaging. Invest Radiol. 2018;53(4):229–35. pmid:29190227
  9. 9. Han L, Zhu Y, Liu Z, Yu T, He C, Jiang W, et al. Radiomic nomogram for prediction of axillary lymph node metastasis in breast cancer. Eur Radiol. 2019;29(7):3820–9. pmid:30701328
  10. 10. Wang Q, Lin Y, Ding C, Guan W, Zhang X, Jia J, et al. Multi-modality radiomics model predicts axillary lymph node metastasis of breast cancer using MRI and mammography. Eur Radiol. 2024. Online ahead of print. pmid:38337068
  11. 11. Valente SA, Levine GM, Silverstein MJ, Rayhanabad JA, Weng-Grumley JG, Ji L, et al. Accuracy of predicting axillary lymph node positivity by physical examination, mammography, ultrasonography, and magnetic resonance imaging. Ann Surg Oncol. 2012;19(6):1825–30. pmid:22227922
  12. 12. The application of radiomics in breast MRI: a review. Technol Cancer Res Treat. 2020;19: 1533033820916191. pmid:32347167
  13. 13. Bi WL, Hosny A, Schabath MB, Giger ML, Birkbak NJ, Mehrtash A, et al. Artificial intelligence in cancer imaging: Clinical challenges and applications. CA Cancer J Clin. 2019;69(2):127–57. pmid:30720861
  14. 14. Gillies RJ, Kinahan PE, Hricak H. Radiomics: Images Are More than Pictures, They Are Data. Radiology. 2016;278(2):563–77. pmid:26579733
  15. 15. Arefan D, Chai R, Sun M, Zuley ML, Wu S. Machine learning prediction of axillary lymph node metastasis in breast cancer: 2D versus 3D radiomic features. Med Phys. 2020;47(12):6334–42. pmid:33058224
  16. 16. Cheng Y, Xu S, Wang H, Wang X, Niu S, Luo Y, et al. Intra- and peri-tumoral radiomics for predicting the sentinel lymph node metastasis in breast cancer based on preoperative mammography and MRI. Front Oncol. 2022;12:1047572. pmid:36578933
  17. 17. Santucci D, Faiella E, Cordelli E, Sicilia R, de Felice C, Zobel BB, et al. 3T MRI-Radiomic Approach to Predict for Lymph Node Status in Breast Cancer Patients. Cancers (Basel). 2021;13(9):2228. pmid:34066451
  18. 18. McInnes MDF, Moher D, Thombs BD, McGrath TA, Bossuyt PM, and the P-DTAG, et al. Preferred Reporting Items for a Systematic Review and Meta-analysis of Diagnostic Test Accuracy Studies: The PRISMA-DTA Statement. JAMA. 2018;319(4):388–96. pmid:29362800
  19. 19. Whiting PF, Rutjes AW, Westwood ME, Mallett S, Deeks JJ, Reitsma JB, et al. QUADAS-2: a revised tool for the quality assessment of diagnostic accuracy studies. Ann Intern Med. 2011;155(8):529–36. pmid:22007046
  20. 20. Kocak BD’Antonoli TA, Mercaldo N, Alberich-Bayarri A, Baessler B, Ambrosini I, et al. METhodological RadiomICs Score (METRICS): a quality scoring tool for radiomics research endorsed by EuSoMII. Insights Imaging. 2024;15(1):8. pmid:38228979
  21. 21. Cui X, Wang N, Zhao Y, Chen S, Li S, Xu M, et al. Preoperative Prediction of Axillary Lymph Node Metastasis in Breast Cancer using Radiomics Features of DCE-MRI. Sci Rep. 2019;9(1):2240. pmid:30783148
  22. 22. Liu C, Ding J, Spuhler K, Gao Y, Serrano Sosa M, Moriarty M, et al. Preoperative prediction of sentinel lymph node metastasis in breast cancer by radiomic signatures from dynamic contrast-enhanced MRI. J Magn Reson Imaging. 2019;49(1):131–40. pmid:30171822
  23. 23. Liu J, Sun D, Chen L, Fang Z, Song W, Guo D, et al. Radiomics Analysis of Dynamic Contrast-Enhanced Magnetic Resonance Imaging for the Prediction of Sentinel Lymph Node Metastasis in Breast Cancer. Front Oncol. 2019;9:980. pmid:31632912
  24. 24. Shan YN, Xu W, Wang R, Wang W, Pang PP, Shen QJ. A Nomogram Combined Radiomics and Kinetic Curve Pattern as Imaging Biomarker for Detecting Metastatic Axillary Lymph Node in Invasive Breast Cancer. Front Oncol. 2020;10:1463. pmid:32983979
  25. 25. Wang C, Chen X, Luo H, Liu Y, Meng R, Wang M, et al. Development and Internal Validation of a Preoperative Prediction Model for Sentinel Lymph Node Status in Breast Cancer: Combining Radiomics Signature and Clinical Factors. Front Oncol. 2021;11:754843. pmid:34820327
  26. 26. Zhan C, Hu Y, Wang X, Liu H, Xia L, Ai T. Prediction of Axillary Lymph Node Metastasis in Breast Cancer using Intra-peritumoral Textural Transition Analysis based on Dynamic Contrast-enhanced Magnetic Resonance Imaging. Acad Radiol. 2022;29 Suppl 1:S107–S15. pmid:33712393
  27. 27. Zhu Y, Yang L, Shen H. Value of the Application of CE-MRI Radiomics and Machine Learning in Preoperative Prediction of Sentinel Lymph Node Metastasis in Breast Cancer. Front Oncol. 2021;11:757111. pmid:34868967
  28. 28. Chen D, Liu X, Hu C, Hao R, Wang O, Xiao Y. Radiomics-based signature of breast cancer on preoperative contrast-enhanced MRI to predict axillary metastasis. Future Oncol. 2022:1–14. pmid:36475996
  29. 29. Chen W, Lin G, Kong C, Wu X, Hu Y, Chen M, et al. Non-invasive prediction model of axillary lymph node status in patients with early-stage breast cancer: a feasibility study based on dynamic contrast-enhanced-MRI radiomics. Br J Radiol. 2024;97(1154):439–50. pmid:38308028
  30. 30. Liu Y, Li X, Zhu L, Zhao Z, Wang T, Zhang X, et al. Preoperative Prediction of Axillary Lymph Node Metastasis in Breast Cancer Based on Intratumoral and Peritumoral DCE-MRI Radiomics Nomogram. Contrast Media Mol Imaging. 2022;2022:6729473. pmid:36051932
  31. 31. Ma M, Jiang Y, Qin N, Zhang X, Zhang Y, Wang X, et al. A Radiomics Model for Preoperative Predicting Sentinel Lymph Node Metastasis in Breast Cancer Based on Dynamic Contrast-Enhanced MRI. Front Oncol. 2022;12:884599. pmid:35734587
  32. 32. Song D, Yang F, Zhang Y, Guo Y, Qu Y, Zhang X, et al. Dynamic contrast-enhanced MRI radiomics nomogram for predicting axillary lymph node metastasis in breast cancer. Cancer Imaging. 2022;22(1):17. pmid:35379339
  33. 33. Tang Y, Chen L, Qiao Y, Li W, Deng R, Liang M. Radiomic Signature Based on Dynamic Contrast-Enhanced MRI for Evaluation of Axillary Lymph Node Metastasis in Breast Cancer. Comput Math Methods Med. 2022;2022:1507125. pmid:36035302
  34. 34. Wang Y, Shang Y, Guo Y, Hai M, Gao Y, Wu Q, et al. Clinical study on the prediction of ALN metastasis based on intratumoral and peritumoral DCE-MRI radiomics and clinico-radiological characteristics in breast cancer. Front Oncol. 2024;14:1357145. pmid:38567148
  35. 35. Zhang J, Zhang Z, Mao N, Zhang H, Gao J, Wang B, et al. Radiomics nomogram for predicting axillary lymph node metastasis in breast cancer based on DCE-MRI: A multicenter study. J Xray Sci Technol. 2023;31(2):247–63. pmid:36744360
  36. 36. Chen JM, Zhu HY, Gao J, Ge YQ, Wang MH, Li Y, et al. Radiomics models based on clinical-pathology and conventional and functional MRI for predicting lymph node metastases of breast cancer axillary. Chin J Med Imaging Technol.2021;37(6):885–90.
  37. 37. Zhang CM, Ding ZM, Chen P, Liu QF. The value of machine learning models for predicting axillary lymph node metastasis in breast cancer based on intratumoral and peritumoral radiomics features of DCE-MRI. Chin Comput Med Imaging, 2023;29(6):618–24.
  38. 38. Zhao NN, Zhu Y, Tang XM, Yang Z, Li Y, Zhang SN, et al. Prediction of axillary lymph node metastasis in breast cancer based on intro-tumoral and peri-tumoral MRI radiomics nomogram. Chin J Megn Reson Imaging, 2023,14(3):81–7.
  39. 39. Zhu YQ, Ji H, Zhu YF, Lv J, Liu Y. Predictive value of preoperative MRI-based nomogram for axillary lymph node metastasis in breast cancer. Chin J Magn Reson Imaging. 2022;13(5):52–8.
  40. 40. Luckett DJ, Laber EB, El-Kamary SS, Fan C, Jhaveri R, Perou CM, et al. Receiver operating characteristic curves and confidence bands for support vector machines. Biometrics. 2021;77(4):1422–30. pmid:32865820
  41. 41. Yu Y, He Z, Ouyang J, Tan Y, Chen Y, Gu Y, et al. Magnetic resonance imaging radiomics predicts preoperative axillary lymph node metastasis to support surgical decisions and is associated with tumor microenvironment in invasive breast cancer: A machine learning, multicenter study. EBioMedicine. 2021;69:103460. pmid:34233259
  42. 42. Yu Y, Tan Y, Xie C, Hu Q, Ouyang J, Chen Y, et al. Development and Validation of a Preoperative Magnetic Resonance Imaging Radiomics-Based Signature to Predict Axillary Lymph Node Metastasis and Disease-Free Survival in Patients With Early-Stage Breast Cancer. JAMA Netw Open. 2020;3(12):e2028086. pmid:33289845
  43. 43. Zhou J, Zhang Y, Chang KT, Lee KE, Wang O, Li J, et al. Diagnosis of Benign and Malignant Breast Lesions on DCE-MRI by Using Radiomics and Deep Learning With Consideration of Peritumor Tissue. J Magn Reson Imaging. 2020;51(3):798–809. pmid:31675151
  44. 44. Braman N, Prasanna P, Whitney J, Singh S, Beig N, Etesami M, et al. Association of Peritumoral Radiomics With Tumor Biology and Pathologic Response to Preoperative Targeted Therapy for HER2 (ERBB2)-Positive Breast Cancer. JAMA Netw Open. 2019;2(4):e192561. pmid:31002322
  45. 45. Braman NM, Etesami M, Prasanna P, Dubchuk C, Gilmore H, Tiwari P, et al. Intratumoral and peritumoral radiomics for the pretreatment prediction of pathological complete response to neoadjuvant chemotherapy based on breast DCE-MRI. Breast Cancer Res. 2017;19(1):57. pmid:28521821
  46. 46. Pineda FD, Medved M, Fan X, Ivancevic MK, Abe H, Shimauchi A, et al. Comparison of dynamic contrast-enhanced MRI parameters of breast lesions at 1.5 and 3.0 T: a pilot study. Br J Radiol. 2015;88(1049):20150021. pmid:25785918
  47. 47. Chen C, Qin Y, Chen H, Zhu D, Gao F, Zhou X. A meta-analysis of the diagnostic performance of machine learning-based MRI in the prediction of axillary lymph node metastasis in breast cancer patients. Insights Imaging. 2021;12(1):156. pmid:34731343
  48. 48. Gong X, Guo Y, Zhu T, Peng X, Xing D, Zhang M. Diagnostic performance of radiomics in predicting axillary lymph node metastasis in breast cancer: A systematic review and meta-analysis. Front Oncol. 2022;12:1046005. pmid:36518318
  49. 49. Lin J, Zheng H, Jia Q, Shi J, Wang S, Wang J, et al. A meta-analysis of MRI radiomics-based diagnosis for BI-RADS 4 breast lesions. J Cancer Res Clin Oncol. 2024;150(5):254. pmid:38748373
  50. 50. Liu CJ, Zhang L, Sun Y, Geng L, Wang R, Shi KM, et al. Application of CT and MRI images based on an artificial intelligence algorithm for predicting lymph node metastasis in breast cancer patients: a meta-analysis. BMC Cancer. 2023;23(1):1134. pmid:37993845
  51. 51. Zhang J, Li L, Zhe X, Tang M, Zhang X, Lei X, et al. The Diagnostic Performance of Machine Learning-Based Radiomics of DCE-MRI in Predicting Axillary Lymph Node Metastasis in Breast Cancer: A Meta-Analysis. Front Oncol. 2022;12:799209. pmid:35186739