Reading time: 5 minutes

Laura Urbina Jara

How AI works in breast cancer detection

Today, it may seem that AI is present in different fields, and the medical field is no exception. Physicians are currently testing AI tools for cancer detection. Artificial intelligence (AI) is a computer technology that learns from data, similar to how we learn from experiences.¹ 

The premise is that AI could be trained to detect breast cancer faster and more precisely. But how is this possible? An AI program analyzes thousands of breast cancer scans, which is the way it ‘learns’. After analyzing lots of images, AI knows what early breast cancer looks like, so it can detect subtle changes or patterns that may indicate cancer.²  

How can AI play a role in detecting early breast cancer cases?

The numbers for breast cancer incidence and mortality are high. In the United States, one in eight women will get breast cancer, and one in 43 will die from the disease.³ This reason is why there is an urgent unmet need to detect breast cancer, and AI can be of additional help in early detection.

There are three image-based modalities used by healthcare providers: mammography, ultrasound, and MRI. Radiologists are the ones who detect breast cancer, even at their best performance, they may still overlook some cancer lesions. Sometimes the workload is high, and human eyes can get tired. AI can still detect suspicious lesions and alert radiologists, reducing their work by 88%⁴. In recent studies, AI has been able to detect breast cancer as well as expert doctors. In some cases, AI showed a higher detection rate (11.5%) than specialists, and AI was able to detect some overlooked cancer suspicious images.^4,5

Approved AI tools 

The Food and Drug Administration (FDA) has approved certain AI programs for breast cancer detection, including mammogram and ultrasound tools.⁶ It is important to note that only trained physicians, such as radiologists, are allowed to use this type of software, and radiologists are not being replaced.

Tools for mammograms:

• Saige-Dx: Detects if there are lesions in mammography images in women over 35.
• MammoScreen® 2.0: scores images. Low scores mean low cancer suspicion, and high scores mean high cancer suspicion. 
• HealthMammo: alerts for suspicious scans so that physicians can prioritize them. 

Tools for breast ultrasounds:

• Koios DS for Breast: classifies regions with breast lesions into four categories–from benign to malignant. 
• BU-CAD: highlights areas that seem suspicious.
• QVCAD System: marks areas that need attention. 

Challenges And Concerns In AI Use

Despite the promising results and controlled use of AI in early breast cancer detection, there are challenges and concerns that need to be addressed. Some concerns include bias, patient privacy, regulation, and responsibility.⁶

These AI models need to be tested across different sites and populations to ensure accuracy and reduce bias while maintaining the protection of intellectual property and patient privacy, according to HIPAA.⁷ Additionally, researchers are testing some possible solutions for these challenges and concerns such as the implementation of centralized learning to ensure all data is collected in one place for AI learning. Patient privacy protection can use de-identified data, so all personal information is removed, and only the data is shared. In order to maintain patients’ privacy, Federated learning is proposed as a solution. In this approach, the data remain private at each center or institution, but the machine learning models are updated together and shared. Ogier du Terrail and colleagues used this approach to develop an AI model that predicts how patients with triple-negative breast cancer will respond to neoadjuvant chemotherapy.^7,8

The main goal of AI use in early breast cancer detection is to be a tool for doctors, not to replace them. In some cases,  AI can be a problem-solving tool for some health centers with difficult access and resources. Interestingly, over 80% of approved AI tools are designed to help cancer detection and diagnosis.⁹

A recent study conducted by Louis and collaborators analyzed the use of AI to improve early breast cancer detection, the AI-Supported Safeguard Review Evaluation (ASSURE) study. They used an AI tool for detection and diagnosis as an initial interpretation, and a breast imaging radiologist performed an additional review of high-risk cases. Results showed a 21.6% increase in cancer detection rate (5.6 versus 4.6 per 1000), suggesting improved screening effectiveness.¹⁰ Similarly, Gommers and collaborators compared the use of AI in mammography screening versus double reading without AI in interval cancers. They suggest that using AI in screening programs can improve breast cancer detection and this tool could be especially useful in places where there is a shortage of trained healthcare personnel.¹¹

Takeaway

Breast cancer is a major public health concern, and its early detection is key to better treatments and outcomes. The use of AI is progressing and could even be considered a tool in breast cancer detection. It is important to note that AI tools are helping doctors, not replacing them. AI can not replace human intuition, empathy, and reasoning. Promising results of AI use in recent studies suggest that the implementations of these tools could be a reality in the future. But before that happens, concerns and challenges such as bias, patient data privacy, regulation, and responsibility should be carefully addressed.

Header Image Source: Created by author with Canva

Edited by Rosa Fontana

References

1.     AI and Cancer – NCI Available online: https://www.cancer.gov/research/infrastructure/artificial-intelligence (accessed on 4 May 2025).

2.     Rentiya, Z.S.; Mandal, S.; Inban, P.; Vempalli, H.; Dabbara, R.; Ali, S.; Kaur, K.; Adegbite, A.; Intsiful, T.A.; Jayan, M.; et al. Revolutionizing Breast Cancer Detection With Artificial Intelligence (AI) in Radiology and Radiation Oncology: A Systematic Review. Cureus 2024, doi:10.7759/cureus.57619.

3.     Breast Cancer Statistics | How Common Is Breast Cancer? Available online: https://www.cancer.org/cancer/types/breast-cancer/about/how-common-is-breast-cancer.html (accessed on 4 May 2025).

4.     Pallumeera, M.; Giang, J.C.; Singh, R.; Pracha, N.S.; Makary, M.S. Evolving and Novel Applications of Artificial Intelligence in Cancer Imaging. Cancers 2025, 17, 1510, doi:10.3390/cancers17091510.

5.     Díaz, O.; Rodríguez-Ruíz, A.; Sechopoulos, I. Artificial Intelligence for Breast Cancer Detection: Technology, Challenges, and Prospects. Eur J Radiol 2024, 175, 111457, doi:10.1016/j.ejrad.2024.111457.

6.     Hunter B, Hindocha S, Lee RW. The Role of Artificial Intelligence in Early Cancer Diagnosis. Cancers (Basel). 2022;14(6):1524. doi:10.3390/cancers14061524

7.     Lotter, W.; Hassett, M.J.; Schultz, N.; Kehl, K.L.; Van Allen, E.M.; Cerami, E. Artificial Intelligence in Oncology: Current Landscape, Challenges, and Future Directions. Cancer Discov 2024, 14, 711–726, doi:10.1158/2159-8290.CD-23-1199.

8.     Ogier du Terrail, J.; Leopold, A.; Joly, C.; Béguier, C.; Andreux, M.; Maussion, C.; Schmauch, B.; Tramel, E.W.; Bendjebbar, E.; Zaslavskiy, M.; et al. Federated Learning for Predicting Histological Response to Neoadjuvant Chemotherapy in Triple-Negative Breast Cancer. Nat Med 2023, 29, 135–146, doi:10.1038/s41591-022-02155-w.

9.    Luchini, C.; Pea, A.; Scarpa, A. Artificial Intelligence in Oncology: Current Applications and Future Perspectives. Br J Cancer 2022, 126, 4–9, doi:10.1038/s41416-021-01633-1.

10.     Louis, L.D.; Wakelin, E.A.; McCabe, M.P.; Ng, A.Y.; Kim, J.G.; Lee, C.I.; Buist, D.S.M.; Gregory Sorensen, A.; Haslam, B. Equitable Impact of an AI-Driven Breast Cancer Screening Workflow in Real-World US-Wide Deployment. Nat. Health 2026, 1, 58–66, doi:10.1038/s44360-025-00001-0.

11.     Gommers, J.; Hernström, V.; Josefsson, V.; Sartor, H.; Schmidt, D.; Hjelmgren, A.; Larsson, A.-M.; Hofvind, S.; Andersson, I.; Rosso, A.; et al. Interval Cancer, Sensitivity, and Specificity Comparing AI-Supported Mammography Screening with Standard Double Reading without AI in the MASAI Study: A Randomised, Controlled, Non-Inferiority, Single-Blinded, Population-Based, Screening-Accuracy Trial. The Lancet 2026, 407, 505–514, doi:10.1016/S0140-6736(25)02464-X.