Guihai Pan , Zejun Pan , Wubiao Chen , Yongjun Wu , Xiaoqing Di , Fei Zhou , Yuting Liao
{"title":"Integration of MRI radiomics and clinical data for preoperative prediction of vascular invasion in breast cancer: A deep learning approach","authors":"Guihai Pan , Zejun Pan , Wubiao Chen , Yongjun Wu , Xiaoqing Di , Fei Zhou , Yuting Liao","doi":"10.1016/j.mri.2025.110339","DOIUrl":null,"url":null,"abstract":"<div><h3>Background</h3><div>Accurate preoperative prediction of vascular invasion in breast cancer is crucial for surgical planning and patient management. MRI radiomics has shown promise in enhancing diagnostic precision. This study aims to evaluate the effectiveness of integrating MRI radiomic features with clinical data using a deep learning approach to predict vascular invasion in breast cancer patients.</div></div><div><h3>Methods</h3><div>A retrospective analysis was conducted on 102 patients with invasive breast cancer confirmed by surgical pathology. Using the MR750 3.0 T as the examination device, the subject underwent the examination in standard breast positions and sequences. Diffusion-weighted imaging (DWI) was performed with two selected b-values, specifically 0 and 1000 s/mm<sup>2</sup>. Following the injection of the contrast agent, dynamic scans were conducted across six phases, and delayed phase sagittal images were acquired using the VIBRANT sequence. Texture features were extracted from MRI images, and key radiomic and clinical features were selected using variance thresholding, correlation filtering, and logistic regression. A predictive model was developed combining these features, and its performance was evaluated through sensitivity, specificity, and area under the curve (AUC) metrics.</div></div><div><h3>Results</h3><div>The univariate models based on individual MRI sequences or clinical data demonstrated variable diagnostic performance. In contrast, the multifactorial model that combined radiomic features with clinical data achieved significantly higher accuracy, with an AUC of 0.829, sensitivity of 76.9 %, and specificity of 83.3 %.</div></div><div><h3>Conclusion</h3><div>Integrating MRI radiomics and clinical data enhances the preoperative prediction of vascular invasion in breast cancer. This approach can improve diagnostic accuracy, providing valuable insights for clinical decision-making and personalized treatment strategies.</div></div>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"118 ","pages":"Article 110339"},"PeriodicalIF":2.1000,"publicationDate":"2025-01-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Magnetic resonance imaging","FirstCategoryId":"3","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0730725X25000219","RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"RADIOLOGY, NUCLEAR MEDICINE & MEDICAL IMAGING","Score":null,"Total":0}
引用次数: 0
Abstract
Background
Accurate preoperative prediction of vascular invasion in breast cancer is crucial for surgical planning and patient management. MRI radiomics has shown promise in enhancing diagnostic precision. This study aims to evaluate the effectiveness of integrating MRI radiomic features with clinical data using a deep learning approach to predict vascular invasion in breast cancer patients.
Methods
A retrospective analysis was conducted on 102 patients with invasive breast cancer confirmed by surgical pathology. Using the MR750 3.0 T as the examination device, the subject underwent the examination in standard breast positions and sequences. Diffusion-weighted imaging (DWI) was performed with two selected b-values, specifically 0 and 1000 s/mm2. Following the injection of the contrast agent, dynamic scans were conducted across six phases, and delayed phase sagittal images were acquired using the VIBRANT sequence. Texture features were extracted from MRI images, and key radiomic and clinical features were selected using variance thresholding, correlation filtering, and logistic regression. A predictive model was developed combining these features, and its performance was evaluated through sensitivity, specificity, and area under the curve (AUC) metrics.
Results
The univariate models based on individual MRI sequences or clinical data demonstrated variable diagnostic performance. In contrast, the multifactorial model that combined radiomic features with clinical data achieved significantly higher accuracy, with an AUC of 0.829, sensitivity of 76.9 %, and specificity of 83.3 %.
Conclusion
Integrating MRI radiomics and clinical data enhances the preoperative prediction of vascular invasion in breast cancer. This approach can improve diagnostic accuracy, providing valuable insights for clinical decision-making and personalized treatment strategies.
期刊介绍:
Magnetic Resonance Imaging (MRI) is the first international multidisciplinary journal encompassing physical, life, and clinical science investigations as they relate to the development and use of magnetic resonance imaging. MRI is dedicated to both basic research, technological innovation and applications, providing a single forum for communication among radiologists, physicists, chemists, biochemists, biologists, engineers, internists, pathologists, physiologists, computer scientists, and mathematicians.