Localizing healthcare: Teaming up to improve breast cancer screening in Vietnam with AI

Breast cancer screening plays a critical role in improving women's health. For example, in Vietnam, breast cancer is the most common cancer among women, yet many do not have access to screening or diagnostic tests. As a result, the disease is often detected late and therefore with a poor prognosis. One reason for this is that there are too few and too few specialized radiologists.

To help address this issue, Siemens Healthineers is part of an international team that uses AI to give more women in Vietnam access to prevention, detect more cancer cases earlier, reduce false positive rates, and thus contribute to a more sustainable healthcare system.

Together with the non-profit aid organization Radiology Across Borders, Hanoi Medical University Hospital, and our strategic partner ScreenPoint Medical, we have created a pilot project to evaluate anonymized mammograms from around 13,500 women. The images are read by medical professionals and at the same time by an AI, and the results are then compared. The AI detects suspicious regions on the X-ray images and alerts the radiologists – providing additional security in a healthcare professional’s daily work. The AI also aims to compensate for differences in experience between radiologists specializing in breast cancer and radiologists with a general education. The algorithm, which has already been approved in the EU and the US, has been trained with numerous mammograms, recognizing suspicious structures, and assigning them a score from 1 (low probability of cancer) to 10 (high probability).

Scientifically, the benefit of AI in breast cancer screening has long been proven: According to the Radiological Society of North America (RSNA), the use of AI can enhance the performance of radiologists by increasing the breast cancer detection rate as compared to manual screening. While AI will not replace radiologists, the use of this technology can help reduce the reading time for each mammography screening, allowing for greater attention to be directed towards suspicious cases, while improving efficiency .

However, the AI that we are evaluating with our partners now in Vietnam has been primarily trained with images from women in the EU and US. It is known that Asian women have a higher breast density, i.e., they have more glandular and connective tissue, and less fatty tissue than women from other parts of the world. This can influence the risk of breast cancer, and the higher breast density also makes it more difficult to detect breast cancer with X-ray. The ongoing research now supports women in Vietnam to also benefit from the automatic image recognition as the AI can be trained with local data. The outcomes of this pilot will hopefully encourage other countries in the region and beyond to expand AI usage for better patient outcomes.

This project is a great example of cross-border teamwork: experts from Vietnam, Australia, New Zealand, the Netherlands, and Germany are involved – a truly global effort to improve access to prevention in previously underserved regions and in the fight against one of the most threatening diseases for women. The challenge is huge: breast cancer is the world’s most prevalent cancer and the leading cause of cancer-related deaths among women worldwide. Approximately eight million women worldwide are living with a diagnosis of breast cancer. In 2020, 2.3 million women were diagnosed with the cancer, and an estimated 685,000 died as a result.

While technology such as AI is important, it is not the sole contribution to better prevention and thus a more sustainable healthcare system. Health is often and decisively co-determined by social determinants, such as education, income, location, gender equality and, last but not least, internet connectivity in order to be able to obtain medical information. If, thanks to the support of AI, healthcare systems can offer screenings to more women at a lower cost, we will be one step closer to our goal of better medical care.

Our long-term objective is to establish automated examination reports, so that radiologists can have more bandwidth to increase their efficiency. We are also working on personalized recommendations for each individual woman. This personalized diagnosis and treatment could minimize radiation exposure and interventions such as biopsies and contribute to better early detection – an important step toward greater precision in medicine.

This joint project of healthcare professionals, a non-profit aid organization, medtech, and AI-experts shows that collaboration and the adaptation of global technology to local conditions can help more people get better and “glocal” access to care – no matter who they are or where they live.

---

October is Breast Cancer Awareness Month, let’s spread the word about early detection and #GetChecked