Siam Tahsin Bhuiyan
- Intern, CCDS
Bachelors
Independent University, Bangladesh
Conference Paper2025
The Role of Model Architecture in Breast Cancer Classification: A Comparison of CNN and Transformer Paradigms on Ultrasound Data
Siam Tahsin Bhuiyan, Halima Khatun, Samiul Karim Mazumder, Fatin Israq, Riyadul Islam, Rashedur Rahman, Ashraful Islam, Saadia Binte Alam
2025 IEEE International Women in Engineering (WIE) Conference on Electrical and Computer Engineering (WIECON-ECE)
In: 2025 IEEE International Women in Engineering (WIE) Conference on Electrical and Computer Engineering (WIECON-ECE)
IEEE, pp. 356-361
Abstract
Breast cancer is the leading malignancy in women globally, and early detection is critical for improving survival. Deep learning has shown promise in medical image analysis, yet the effects of model architecture remain underexplored. This study conducts a comparative evaluation of four models: DenseNet121 (pure CNN), ConvNeXt (modernized CNN with transformer-inspired design), ViT-B/16 (pure Vision Transformer), and Swin-B (hierarchical transformer with CNN-like inductive biases) for breast ultrasound classification using the public BUSI dataset of 780 images. Results demonstrate that DenseNet121 exhibits limitations due to its reliance on local receptive fields, which restricts its ability to capture distributed pathological features. ConvNeXt outperforms DenseNet121 across all metrics, with 83% accuracy and 91% AUROC, highlighting the benefits of transformer-inspired refinements within CNNs. ViT-B/16 achieves the highest AUROC of 94%, underscoring the strength of global self-attention for capturing distributed pathological cues. Swin-B attains the most balanced performance, with 86% accuracy, 85% recall, and 0.92 AUROC, reflecting the effectiveness of combining global attention with hierarchical locality. These findings suggest a clear performance hierarchy, where hybrid transformer architectures such as Swin-B offer the most clinically reliable balance between sensitivity and specificity. Future research should explore hybrid models that integrate CNN inductive priors with transformer-based global reasoning to further enhance robustness in real-world diagnostic applications.
Keywords
Breast cancerUltrasoundArchitectureTransformerPattern recognition (psychology)
CCDS Authors
Halima Khatun
- Research Assistant, CCDS
Bachelors
Independent University, Bangladesh
Samiul Karim Mazumder
- Intern, CCDS
Riyadul Islam
- Intern, CCDS
Bachelors
Daffodil International University
Md. Rashedur Rahman, D. Eng.
- Co-Director, MIRA & NEST Wings
Assistant Professor
Department of Computer Science and Engineering
Independent University, Bangladesh
Ashraful Islam, PhD
- Center Director, CCDS
- Director, HCI Wing
Assistant Professor
Department of Computer Science and Engineering
Independent University, Bangladesh (IUB)Research Interests
Human-Computer Interaction, AI for Social Good, AI for Public Health, AI for Impact
Saadia Binte Alam, PhD
- Director, MIRA & WiSE Wings
Associate Professor
Department of Computer Science and Engineering
Independent University, Bangladesh
References
- 1.Ze Liu, Yutong Lin, Yue Cao, Han Hu, Yixuan Wei, Zheng Zhang, Stephen Lin, Baining Guo. (2021). Swin Transformer: Hierarchical Vision Transformer using Shifted Windows. 2021 IEEE/CVF International Conference on Computer Vision (ICCV), 9992–10002[10.1109/iccv48922.2021.00986]
- 2.Zhuang Liu, Hanzi Mao, Chao-Yuan Wu, Christoph Feichtenhofer, Trevor Darrell, Saining Xie. (2022). A ConvNet for the 2020s. 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 11966–11976[10.1109/cvpr52688.2022.01167]
- 3.Wenya Linda Bi, Ahmed Hosny, Matthew B. Schabath, Maryellen L. Giger, Nicolai J. Birkbak, Alireza Mehrtash, Tavis Allison, Omar Arnaout, Christopher Abbosh, Ian F. Dunn, Raymond H. Mak, Rulla M. Tamimi, Clare M. Tempany, Charles Swanton, Udo Hoffmann, Lawrence H. Schwartz, Robert J. Gillies, Raymond Y. Huang, Hugo J. W. L. Aerts. (2019). Artificial intelligence in cancer imaging: Clinical challenges and applications. CA A Cancer Journal for Clinicians, 69(2), 127–157[10.3322/caac.21552]
- 4.Béatrice Lauby-Secretan, Chiara Scoccianti, Dana Loomis, Lamia Benbrahim-Tallaa, Véronique Bouvard, Franca Bianchini, Kurt Straif. (2015). Breast-Cancer Screening — Viewpoint of the IARC Working Group. New England Journal of Medicine, 372(24), 2353–2358[10.1056/nejmsr1504363]
- 5.Evan R. Myers, Patricia Moorman, Jennifer M. Gierisch, Laura J. Havrilesky, Lars J. Grimm, Sujata Ghate, Brittany Davidson, Ranee Chatterjee Mongtomery, Matthew J. Crowley, Douglas C. McCrory, Amy Kendrick, Gillian D. Sanders. (2015). Benefits and Harms of Breast Cancer Screening. JAMA, 314(15), 1615[10.1001/jama.2015.13183]
- 6.Woo Kyung Moon, Yan-Wei Lee, Hao-Hsiang Ke, Su Hyun Lee, Chiun-Sheng Huang, Ruey-Feng Chang. (2020). Computer‐aided diagnosis of breast ultrasound images using ensemble learning from convolutional neural networks. Computer Methods and Programs in Biomedicine, 190, 105361[10.1016/j.cmpb.2020.105361]
- 7.Richard Adam, Kevin Dell’Aquila, Laura Hodges, Takouhie Maldjian, Tim Q. Duong. (2023). Deep learning applications to breast cancer detection by magnetic resonance imaging: a literature review. Breast Cancer Research, 25(1), 87[10.1186/s13058-023-01687-4]
- 8.Jung Hyun Yoon, Eun-Kyung Kim. (2021). Deep Learning-Based Artificial Intelligence for Mammography. Korean Journal of Radiology, 22(8), 1225[10.3348/kjr.2020.1210]
- 9.Adyasha Sahu, Pradeep Kumar Das, Sukadev Meher. (2023). Recent advancements in machine learning and deep learning-based breast cancer detection using mammograms. Physica Medica, 114, 103138[10.1016/j.ejmp.2023.103138]
- 10.Mansoor Hayat, Nouman Ahmad, Anam Nasir, Zeeshan Ahmad Tariq. (2024). Hybrid Deep Learning EfficientNetV2 and Vision Transformer (EffNetV2-ViT) Model for Breast Cancer Histopathological Image Classification. IEEE Access, 12, 184119–184131[10.1109/access.2024.3503413]
- 11.M. Latha, P. Santhosh Kumar, R. Roopa Chandrika, T. R. Mahesh, V. Vinoth Kumar, Suresh Guluwadi. (2024). Revolutionizing breast ultrasound diagnostics with EfficientNet-B7 and Explainable AI. BMC Medical Imaging, 24(1), 230[10.1186/s12880-024-01404-3]
- 12.Qing Dan, Tingting Zheng, Li Liu, Desheng Sun, Yun Chen. (2023). Ultrasound for Breast Cancer Screening in Resource-Limited Settings: Current Practice and Future Directions. Cancers, 15(7), 2112[10.3390/cancers15072112]
- 13.Qing Dan, Ziting Xu, Hannah Burrows, Jennifer Bissram, Jeffrey S. A. Stringer, Yingjia Li. (2024). Diagnostic performance of deep learning in ultrasound diagnosis of breast cancer: a systematic review. npj Precision Oncology, 8(1), 21[10.1038/s41698-024-00514-z]
- 14.Di Zhang, Fan Jiang, Rui Yin, Ge-Ge Wu, Qi Wei, Xin-Wu Cui, Shu-E Zeng, Xue-Jun Ni, Christoph F. Dietrich. (2021). A Review of the Role of the S-Detect Computer-Aided Diagnostic Ultrasound System in the Evaluation of Benign and Malignant Breast and Thyroid Masses. Medical Science Monitor, 27, e931957[10.12659/msm.931957]
- 15.Swati Shilaskar, Shripad Bhatlawande, Mayur Talewar, Sidhesh Goud, Soham Tak, Sachi Kurian, Anjali Solanke. (2025). Classification and Segmentation of Breast Tumor Ultrasound Images using VGG-16 and UNet. Biomedical & Pharmacology Journal, 18(1), 569–580[10.13005/bpj/3109]
- 16.Emmanuel Ahishakiye, Fredrick Kanobe. (2025). Breast cancer classification using breast ultrasound images with a hybrid of transfer learning and Bayesian-optimized fast learning network. Discover Artificial Intelligence, 5(1)[10.1007/s44163-025-00335-4]
- 17.S. Sree Kumar, A.K. Radhakrish, S.K. Cheong. (2010). Rapid Metastasis of Breast Cancer Cells from Primary Tumour to Liver. Pakistan Journal of Biological Sciences, 13(7), 303–315[10.3923/pjbs.2010.303.315]
- 18.Walid Al-Dhabyani, Mohammed Gomaa, Hussien Khaled, Aly Fahmy. (2019). Dataset of breast ultrasound images. Data in Brief, 28, 104863[10.1016/j.dib.2019.104863]