Mammography and AI: A Better Tomorrow

Breast cancer is the most common non-skin cancer among women, and the second most deadly. With this significant human toll comes tragedy and cost, estimated at $180 billion in health care expenses and lost productivity per year. Researchers at Google have developed a mammogram-screening tool using AI technology that can reduce cost, increase accuracy, and save lives – and when it came time to choose clinical partners, Google looked to Northwestern.

“Northwestern is an ideal partner for Google because we are one of the largest breast imaging centers in the country,” explains Dr. Sarah Friedewald, Northwestern’s Vice Chair of Women’s Imaging and clinical point-person for the Google-Northwestern collaboration. “We also serve a very diverse population, pulling in women from all across Chicago.” The AI technology uses existing mammogram data to ‘learn’ suspicious patterns within the mammograms. The program can then ‘read’ new mammograms by looking for similar patterns. The more data, the more accurate the predictions. “The AI pre-reads the screening mammograms. A radiologist is immediately notified of any suspicious finding, and then subsequent workup – more imaging or a biopsy – can be performed before the patient leaves the office. This is smarter, faster, and avoids recalls.” Recalls result in excess anxiety and delays in treatment. Approximately 9% of mammograms result in recalls.

Recruitment for the study is ongoing, but publication of results could come as early as next year. “So far, most women have been excited to take part. They know how important screening is, and this trial offers a chance to help women everywhere.” While this project focuses on breast cancer, Dr. Friedewald believes similar projects will soon become common throughout radiology: “Mammograms, CTs, MRIs – all of them are data. AI uses data, and in the future we will use AI to transform patient care. This is just the beginning.”