Back to all papers

A hybrid deep learning framework for accurate breast cancer classification using MRI images.

May 20, 2026pubmed logopapers

Authors

Bakay MS

Affiliations (2)

  • Department of Biomedical Engineering, Faculty of Engineering, Düzce University, Düzce, 81620, Türkiye. [email protected].
  • California State University, State University Drive 5151, Los Angeles, CA, 90032, USA. [email protected].

Abstract

Early and accurate detection of breast cancer is crucial to enhance patient results, especially in high-risk populations where magnetic resonance imaging (MRI) is intensively used. This study shows a deep learning-based framework for the automatic classification of benign and malignant breast lesions using MRI images. To guarantee accurate patient-level annotations, a large-scale dataset comprising around 25,000 breast MRI images was constructed using an extensive preprocessing and labeling procedure. A hybrid feature-fusion model that integrates MobileNetV2 and VGG16 in a parallel structure was extensively compared and examined with a number of transfer learning-based convolutional neural network designs, such as VGG16, MobileNetV2, and DenseNet121. Each model was trained utilizing a two-stage approach that included frozen training and fine-tuning after being initiated with ImageNet pre-trained weights. Accuracy, precision, recall, specificity, F1-score, and ROC-AUC metrics were used to evaluate the model's performance. Having 97.79% accuracy, 96.62% recall, and a ROC-AUC value of 0.9983, the hybrid MobileNetV2-VGG16 model topped all single-model architectures, according to experimental results. The results show that hybrid transfer learning approaches can offer a potential option for computer-aided diagnosis systems and more consistent and reliable classification performance for breast MRI-based cancer detection.

Topics

Journal Article

Ready to Sharpen Your Edge?

Subscribe to join 11k+ peers who rely on RadAI Slice. Get the essential weekly briefing that empowers you to navigate the future of radiology.

We respect your privacy. Unsubscribe at any time.