[공지사항]을 빙자한 안부와 근황 
Show more

Daily Arxiv

This is a page that curates AI-related papers published worldwide.
All content here is summarized using Google Gemini and operated on a non-profit basis.
Copyright for each paper belongs to the authors and their institutions; please make sure to credit the source when sharing.

Enhancing Breast Cancer Detection with Vision Transformers and Graph Neural Networks

Created by
  • Haebom

Author

Yeming Cai, Zhenglin Li, Yang Wang

Outline

This paper presents an innovative framework for early detection of breast cancer. By integrating Vision Transformer (ViT) and Graph Neural Network (GNN), we improve the breast cancer detection accuracy up to 84.2% using the CBIS-DDSM dataset. ViT models global image features, and GNN models structural relationships, achieving better performance than existing methods, and supporting physicians' clinical judgment through interpretable attention heatmaps.

Takeaways, Limitations

•
Takeaways:
◦
Possibility of improving breast cancer detection performance through integration of ViT and GNN.
◦
Presenting the possibility of assisting physician diagnosis using interpretable attention heatmaps.
◦
Achieved improved accuracy (84.2%) compared to existing methods.
•
Limitations:
◦
Further validation is needed to determine whether the presented accuracy (84.2%) fully reflects performance in real clinical settings.
◦
Further study on generalization performance is needed using only the CBIS-DDSM dataset.
◦
There is a need to provide objective evaluation criteria for the interpretability of attention heatmaps.
👍
Made with Slashpage