Comparative evaluation of vision transformers and convolutional networks for breast ultrasound image classification

Abstract

Aim: Interobserver variability continues to limit the consistency of breast ultrasound interpretation. This study compares two Vision Transformer (ViT) models and two Convolutional Neural Network (CNN) models for automated three-class breast ultrasound classification, with a specific focus on the tradeoff between predictive performance and computational efficiency.

Methods: Swin Transformer Base and DeiT Base were evaluated alongside InceptionV3 and MobileNetV3 Large using the public Breast Ultrasound Images (BUSI) dataset, which contains 780 images labeled as benign, malignant, and normal. A consistent on-the-fly augmentation pipeline was applied during training to promote robustness and reduce sensitivity to incidental image variations.

Results: Swin Transformer Base achieved the highest test accuracy (0.9167) and F1 score (0.8981). MobileNetV3 Large reached an accuracy of 0.8583 with substantially lower computational demand. The efficiency contrast was pronounced, with Swin requiring 30.33 GFLOPs versus 0.43 GFLOPs for MobileNetV3 Large.

Conclusions: On this benchmark, ViT models can yield higher classification performance, while lightweight CNNs offer a strong efficiency profile that may better match deployment-constrained settings. These results suggest that model selection should be guided by both predictive accuracy and operational feasibility within the target clinical workflow.

Keywords

breast cancer / deep learning / ultrasound images / computer-aided diagnosis

References

[1] Kim J, Harper A, McCormack V, Sung H, Houssami N, Morgan E, et al. Global patterns and trends in breast cancer incidence and mortality across 185 countries. Nat Med. 2025; 31: 1154-62.

[2] Xiong X, Zheng LW, Ding Y, Chen YF, Cai YW, Wang LP, et al. Breast cancer: pathogenesis and treatments. Sig Transduct Target Ther. 2025; 10: 49.

[3] Obeagu EI, Obeagu GU. Breast cancer: A review of risk factors and diagnosis. Medicine. 2024; 103: e36905.

[4] Kiani P, Vatankhahan H, Zare-Hoseinabadi A, Ferdosi F, Ehtiati S, Heidari P, et al. Electrochemical biosensors for early detection of breast cancer. Clinica Chimica Acta. 2025; 564: 119923.

[5] Alshawwa IA, Qasim El-Mashharawi H, Salman FM, Naji M, Al-Qumboz A, Abu-Nasser BS, et al. Advancements in Early Detection of Breast Cancer: Innovations and Future Directions. Int J Acad Eng Res. 2024; 8: 15-24.

[6] Begum MM, Gupta R, Sunny B, Lutfor ZL. Advancements in Early Detection and Targeted Therapies for Breast Cancer; A Comprehensive Analysis. Asia Pac J Cancer Res. 2024; 1: 4-13.

[7] Siegel RL, Kratzer TB, Giaquinto AN, Sung H, Jemal A. Cancer statistics, 2025. CA Cancer J Clin. 2025; 75: 10-45.

[8] Pacal I, Attallah O. InceptionNeXt-Transformer: A novel multi-scale deep feature learning architecture for multimodal breast cancer diagnosis. Biomed Signal Process Control. 2025; 110: 108116.

[9] Katsika L, Boureka E, Kalogiannidis I, Tsakiridis I, Tirodimos I, Lallas K, et al. Screening for Breast Cancer: A Comparative Review of Guidelines. Life. 2024; 14: 777.

[10] Trentham-Dietz A, Chapman CH, Jayasekera J, Lowry KP, Heckman-Stoddard BM, Hampton JM, et al. Collaborative Modeling to Compare Different Breast Cancer Screening Strategies: A Decision Analysis for the US Preventive Services Task Force. JAMA. 2024; 331: 1947-60.

[11] Cakmak Y, Zeynalov J. A Comparative Analysis of Convolutional Neural Network Architectures for Breast Cancer Classification from Mammograms. Artif Intell Appl Sci. 2025; 1: 28-34.

[12] Cakmak Y, Pacal N. Deep Learning for Automated Breast Cancer Detection in Ultrasound: A Comparative Study of Four CNN Architectures. Artif Intell Appl Sci. 2025; 1: 13-9.

[13] Abu Abeelh E, AbuAbeileh Z. Comparative Effectiveness of Mammography, Ultrasound, and MRI in the Detection of Breast Carcinoma in Dense Breast Tissue: A Systematic Review. Cureus. 2024; 16: e59054.

[14] Pacal I. Chaotic Learning Rate Scheduling for Improved CNN-Based Breast Cancer Ultrasound Classification. Chaos Theory Appl. 2025; 7: 297-306.

[15] Cakmak Y, Pacal I. Comparative analysis of transformer architectures for brain tumor classification. Explor Med. 2025; 6: 1001377.

[16] Iacob R, Iacob ER, Stoicescu ER, Ghenciu DM, Cocolea DM, Constantinescu A, et al. Evaluating the Role of Breast Ultrasound in Early Detection of Breast Cancer in Low- and Middle-Income Countries: A Comprehensive Narrative Review. Bioengineering. 2024; 11: 262.

[17] Gordon PB, Warren LJ, Seely JM. Cancers Detected on Supplemental Breast Ultrasound in Women With Dense Breasts: Update From a Canadian Centre. Can Assoc Radiol J. 2025; 76: 497-507.

[18] Vogel-Minea CM, Bader W, Blohmer JU, Duda V, Eichler C, Fallenberg E, et al. Best Practice Guidelines-DEGUM Recommendations on Breast Ultrasound. Ultraschall Med. 2025; 46: 245-58.

[19] Rana AS, Rafique J, Riffat H. Advances in Breast Ultrasound Imaging: Enhancing Diagnostic Precision and Clinical Utility. London: IntechOpen; 2024.

[20] Pacal I, Algarni A, Kunduracioglu I. Adaptive and efficient deep learning model for automated ischemic stroke lesion segmentation. Biomed Signal Process Control. 2026; 118: 109658.

[21] Cakmak Y. Machine Learning Approaches for Enhanced Diagnosis of Hematological Disorders. Comput Syst Artif Intell. 2025; 1: 8-14.

[22] Çakmak Y, Maman A. Deep Learning for Early Diagnosis of Lung Cancer. Comput Syst Artif Intell. 2025; 1: 20-5.

[23] Pacal I. MaxCerVixT: A novel lightweight vision transformer-based Approach for precise cervical cancer detection. Knowl-Based Syst. 2024; 289: 111482.

[24] Pacal I. Investigating deep learning approaches for cervical cancer diagnosis: a focus on modern image-based models. Eur J Gynaecol Oncol. 2025; 46: 125-41.

[25] Katayama A, Aoki Y, Watanabe Y, Horiguchi J, Rakha EA, Oyama T. Current status and prospects of artificial intelligence in breast cancer pathology: convolutional neural networks to prospective Vision Transformers. Int J Clin Oncol. 2024; 29: 1648-68.

[26] ABIMOULOUD ML, BENSID K, Elleuch M, Ben Ammar M, KHERALLAH M. Vision transformer-based convolutional neural network for breast cancer histopathological images classification. Multimed Tools Appl. 2024; 83: 86833-68.

[27] Balaha HM, Ali KM, Gondim D, Ghazal M, El-Baz A. Harnessing Vision Transformers for Precise and Explainable Breast Cancer Diagnosis. Cham: Springer; 2025. pp. 191-206.

[28] Jahan I, Chowdhury MEH, Vranic S, Al Saady RM, Kabir S, Pranto ZH, et al. Deep learning and vision transformers-based framework for breast cancer and subtype identification. Neural Comput Appl. 2025; 37: 9311-30.

[29] Rahujo A, Atif D, Inam SA, Khan AA, Ullah S. A survey on the applications of transfer learning to enhance the performance of large language models in healthcare systems. Discov Artif Intell. 2025; 5: 90.

[30] Hussain S, Xi X, Ullah I, Inam SA, Naz F, Shaheed K, et al. A Discriminative Level Set Method with Deep Supervision for Breast Tumor Segmentation. Comput Biol Med. 2022; 149: 105995.

[31] Khuhro MA, Inam SA, Iqbal D, Hashim H. An Empirical Approach Towards Detection of Tuberculosis Using Deep Convolutional Neural Network. Int J Data Min Model Manag. 2024; 16: 101-12.

[32] Li M, Fang Y, Shao J, Jiang Y, Xu G, Cui X, et al. Vision transformer-based multimodal fusion network for classification of tumor malignancy on breast ultrasound: A retrospective multicenter study. Int J Med Inform. 2025; 196: 105793.

[33] Duong LT, Doan TTH, Bui AMT, Nguyen PT. Recognition of breast cancer from heterogeneous ultrasound images: A multi-level deep learning approach. Inform Med Unlocked. 2025; 59: 101698.

[34] Li H, Cheng T. Multicenter and multimodal ultrasound-based radiomics and transformer-driven end-to-end deep learning for breast cancer molecular subtype classification. J Radiat Res Appl Sci. 2025; 18: 101656.

[35] Breast Ultrasound Images Dataset(BUSI) [Internet]. [cited 2025 Jun 7]. Available from: https://www.kaggle.com/datasets/sabahesaraki/breast-ultrasound-images-dataset

[36] Wang Z, Wang P, Liu K, Wang P, Fu Y, Lu CT, et al. A Comprehensive Survey on Data Augmentation. arXiv:2405.09591 [Preprint]. 2024 [cited 2025 May 28]. Available from: https://arxiv.org/abs/2405.09591v3

[37] Mumuni A, Mumuni F, Gerrar NK. A Survey of Synthetic Data Augmentation Methods in Machine Vision. Mach Intell Res. 2024; 21: 831-69.

[38] Szegedy C, Vanhoucke V, Ioffe S, Shlens J. Rethinking the Inception Architecture for Computer Vision. CVPR. 2016. pp. 2818-26.

[39] Howard A, Sandler M, Chu G, Chen LC, Chen B, Tan M, et al. Searching for MobileNetV3. Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV). 2019. pp. 1314-24.

[40] Liu Z, Lin Y, Cao Y, Hu H, Wei Y, Zhang Z, et al. Swin Transformer: Hierarchical Vision Transformer using Shifted Windows. Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV). 2021. pp. 10012-22.

[41] Touvron H, Cord M, Douze M, Massa F, Sablayrolles A, Jégou H. Training data-efficient image transformers & distillation through attention. Proc Mach Learn Res. 2021; 139: 10347-57.