Article Sidebar

Published: 2026-08-14
Abstract Views: 0
DOI: https://doi.org/10.52163/yhc.v67iCD12.6154
Issue
Tập 67 Số CĐ12- HNKH Bệnh viện Ung bướu Nghệ An
Section
Articles
How to Cite
Hùng, P. V., Trung, N. Q., Đạt, N. T. B., Tài, H. Đức, Việt, N. V., Trà, T. T. M., Trung, N. B., Nam, N. H., Khoa, T. A., Lâm, P. Đức, & Tình, V. P. (2026). RESEARCH AND APPLICATION OF ARTIFICIAL INTELLIGENCE AND COMPUTER VISION TECHNIQUES FOR ASSISTING IN EARLY DETECTION AND DIAGNOSIS OF BREAST CANCER AT NGHE AN ONCOLOGY HOSPITAL. Vietnam Journal of Community Medicine, 67(CD12). https://doi.org/10.52163/yhc.v67iCD12.6154

RESEARCH AND APPLICATION OF ARTIFICIAL INTELLIGENCE AND COMPUTER VISION TECHNIQUES FOR ASSISTING IN EARLY DETECTION AND DIAGNOSIS OF BREAST CANCER AT NGHE AN ONCOLOGY HOSPITAL

Pham Vinh Hung1, Nguyen Quang Trung1, Nguyen Tai Bui Dat1, Ho Duc Tai1, Nguyen Van Viet1, Tran Thi My Tra1, Nguyen Ba Trung1, Nguyen Hoang Nam2, Tran Anh Khoa2, Pham Duc Lam3, Vo Phuc Tinh2
1 Nghe An Oncology Hospital
2 Ton Duc Thang University
3 Nguyen Tat Thanh University

Main Article Content

Abstract

Objectives: To evaluate the efficiency of artificial intelligence (AI) software version 1.2.1 (combining YOLOv7 and ResNet50 algorithms) in assisting the diagnosis, early detection, localization, and classification of breast cancer lesions on mammograms at Nghe An Oncology Hospital.


Methods: The study was conducted on 400 DICOM-format mammograms from 200 retrospective patients (from March 2024 to August 2024) and 200 prospective patients (from September 2024 to December 2024). The YOLOv7 model was utilized for lesion detection, while ResNet50 was applied for the detailed classification of masses and microcalcifications. The AI diagnostic results were compared against the histopathological gold standard.


Results: In the retrospective phase, the software achieved an accuracy of 94% (188/200 cases) and 100% sensitivity for malignant cases (166/166 cases). In the prospective phase, accuracy reached 91% (182/200 cases) while maintaining 100% sensitivity by correctly identifying all 10 malignant cases, with no critical lesions missed. The inference speed reached 3.6 ms/image. Conclusion: The AI software operates stably, demonstrates high sensitivity, and integrates seamlessly with the PACS system, providing highly effective support for the early detection and screening of breast cancer in clinical practice.

Article Details

Keywords

Breast cancer; Artificial intelligence; Computer vision; YOLOv7; ResNet50; Mammography.

References

1. Đào Văn Tú và cộng sự (2021). Tổng quan nghiên cứu ứng dụng trí tuệ nhân tạo trong chẩn đoán ung thư vú bằng ảnh giải phẫu bệnh kỹ thuật số. Tạp chí Y học Việt Nam, 500(1), tr. 15-20. DOI: https://doi.org/10.51298/vmj.v500i1.290
2. Đặng Công Thuận (2012). Nghiên cứu các đặc điểm chẩn đoán hình ảnh, giải phẫu bệnh và tình trạng thụ thể nội tiết bệnh ung thư vú tại Bệnh viện Trường Đại học Y Dược Huế. Tạp Chí Phụ sản, 10(3), 250-257. DOI: https://doi.org/10.46755/vjog.2012.3.172
3. Nguyễn Thị Mai Lan (2016). Nghiên cứu tỷ lệ mắc mới ung thư vú ở phụ nữ Hà Nội giai đoạn 2014-2016. Luận án tiến sĩ Y học, Đại học Y Hà Nội. URL: http://dulieuso.hmu.edu.vn/handle/hmu/1972
4. Min, H., Wilson, D., Huang, Y., Liu, S., Crozier, S., Bradley, A. P., & Chandra, S. S. (2020). Fully automatic computer-aided mass detection and segmentation via pseudo-color mammograms and mask r-cnn. IEEE 17th International Symposium on Biomedical Imaging (ISBI), pp. 1137-1141 . DOI: https://doi.org/10.1109/ISBI45749.2020.9098312
5. Raza, S. K., Sarwar, S. S., Syed, S. M., & Khan, N. A. (2021). Classification and Segmentation of Breast Tumor Using Mask R-CNN on Mammograms. NED UniversityResearch Journal, 18(4), pp. 79-88. DOI: https://doi.org/10.21203/rs.3.rs-523546/v1