HYBRID CNN-TRANSFORMER FRAMEWORK FOR ACCURATE BREAST TUMOR DETECTION FROM ULTRASOUND IMAGE

The proposed hybrid CNN–Transformer framework for breast tumor detection from ultrasound images provides an efficient, reliable, and automated diagnosis system. By combining CNN’s local feature extraction with Transformer’s global attention mechanism, the model achieves improved classification accuracy and robust detection. The web-based deployment allows real-time tumor prediction, assisting medical professionals in early breast cancer detection and reducing diagnostic errors.   Future enhancements include: ·        Integration with hospital PACS systems ·        Support for multi-modal imaging ·        Mobile application development for wider accessibility.

1.      Dosovitskiy et al., “An Image is      Worth          16x16 Words: Transformers for Image Recognition at Scale,” arXiv, 2020. 2.      “BUSI Breast Ultrasound Dataset,” Public Dataset for Breast Ultrasound Image Analysis   3.      World Health Organization, Breast Cancer Statistics Report, 2024.   4.      Singh, S. P. Mishra, P. Singh, A. Srivastava, “VISNET: An Efficient Light Weighted Hybrid Model for Early Detection of Breast Tumor in Ultrasound Images using Vision Transformer and CNN,” Journal of Information Systems Engineering and Management, Vol. 9 No. 4s, 2024.   5.      Q. He, Q. Yang, M. Xie, “HCTNet: A Hybrid CNN-Transformer Network for Breast Ultrasound Image Segmentation,” Computers in Biology and Medicine, 2023.     6.      M. A. Rahman, “HyFormer-Net: A Synergistic CNN-Transformer with Multi-Scale Fusion for Breast Lesion Segmentation and Classification,” arXiv, Nov 2025.   7.      G. Husain Abbas, S. U. Rehman, G. Bargshady, “A Hybrid CNN-Transformer Model for Multi View Ultrasound Breast Tumor Classification,” medRxiv, August 2025.   8.      Y. Wei, “Breast Ultrasound Image BI-RADS Classification Based on Vision Transformer,” International Journal of Multiphysics, Vol. 18, No. 2, 2024.     9.      B. Maistry, A. E. Ezugwu, “Breast Cancer Detection and Diagnosis: A Comparative Study of Deep Learning Architectures,” arXiv, May 2023.   10.   J. Shi, A. Vakanski, M. Xian, et al., “EMT-NET: Efficient Multitask Network for Breast Cancer Diagnosis,” arXiv, 2022.