Breast cancer is one of the leading causes of death amongst women worldwide and overall, the second leading cause of death after lung cancer. Between 10% and 30% of breast cancer cases can be missed due to factors such as poor positioning, perception errors, and various other reasons. In addition, efforts to reduce the occurrence of false negatives does lead to excessive recalls, where it was shown that 41% of radiologists in the USA have higher recall rates than recommended, and only 28.6% of patients receiving a biopsy are diagnosed with breast cancer.
Advances in deep neural networks and AI that is integrated into breast cancer screening workflows promise to help radiologists identify breast cancer earlier and more consistently improving radiologists' performance and efficiency, as well as increasing workflow efficiency.

High level objectives:

• Develop and deploy AI based binary classification system (normal vs abnormal), to classify mammograms based on the probability of detecting breast tumor.

• Provide a minimum accuracy rate with a sensitivity of 85% and a specificity of 80%, surpassing the radiologists' average screening accuracy of 80%. Several studies have reported that the rate of missed cases by radiologists can reach up to 30%.

• Provide a system that can be integrated to populate AI model result on hospitals PACS.

https://sdaia.gov.sa/en/MediaCenter/News/Pages/NewsDetails.aspx?NewsID=223

• Saudi Arabia

• Women
• Primary Care Centers, Breast Cancer Screening Programs & Radiologists.

This project's AI breast cancer solution boasts impressive replicability due to its carefully constructed modular architecture and inherent data independence. This versatility translates into seamless integration with diverse healthcare infrastructures, enabling customized implementations across data landscapes and resource limitations. By prioritizing interpretability and explainability, the project fosters trust and encourages widespread adoption in various healthcare environments. Consequently, the solution transcends specific platforms or frameworks, becoming a versatile and readily transferable tool in the global fight against breast cancer. This dedication to inclusivity and adaptability empowers the project to reach populations regardless of resource constraints or existing systems, ultimately widening the net of hope in the fight against this critical disease.

Regularly update and maintain AI systems to extend their lifecycle and write clean and efficient code to reduce computational overhead. Regularly report sustainability metrics to stakeholders and make necessary adjustments.

This project has far-reaching effects on healthcare, economic efficiency, and social well-being in Saudi Arabia. By integrating AI technology into breast cancer screening, it directly supports Vision 2030 goals and WSIS values, while addressing critical health challenges 1. WSIS Action Line C3: Access to Information and Knowledge - AI-powered digital mammogram reading expands access to screening, especially for underserved areas. - Reduces late detection rates (currently >55%) by making early screening more efficient and scalable. - Encourages data-driven clinical decision-making by supporting radiologists with AI insights. - Uses deep learning on large datasets annotated by experts, ensuring high diagnostic accuracy. 2. WSIS Action Line C7: ICT Applications – E-Health - Directly supports the National Program for Early Detection of Breast Cancer (established in 2012). - Aligns with KSA’s Vision 2030 Digital Health Strategy to enhance medical services through AI. - Increases radiologists’ efficiency, enabling them to screen more cases in less time. - Reduces false positives and unnecessary referrals, leading to cost savings in the healthcare system. 3. WSIS Action Line C4: Capacity Building - Trains radiologists and medical staff to work with AI-assisted diagnostics. - Strengthens AI research and development in KSA’s healthtech sector. - Engages the healthcare ecosystem, including government bodies, hospitals, and AI researchers. - Multiple workshops conducted to drive innovation and improve healthcare access. 4. WSIS Action Line C5: Building Confidence and Security in ICTs - Uses secure AI models trained on locally sourced mammogram datasets. - Supports cloud-based AI solutions while maintaining patient data privacy. - AI-assisted early detection increases confidence in medical technology. - Reduces reliance on manual readings, decreasing human error rates in diagnosis.