How Artificial Intelligence Will Improve Early Breast Cancer Detection
In the United States alone, breast cancer claims over 40,000 women’s lives every year. Yet, the problem is not exactly the cancer itself, but rather the stage during which it is detected. When a cancer is detected at an early stage, it can often be cured. Although mammograms are the most effective tools for detecting breast cancer available to date, they are still flawed and they can present false-positive results. This, in turn, leads to unnecessary biopsies and surgical procedures.
One of the most common causes of false-positive results are high-risk lesions. These lesions appear suspicious on mammograms because they portray abnormal cells. A patient typically undergoes a surgical procedure to have the lesions removed, but 90% of the time, the lesions turn out to be benign. As a result, thousands of women undergo painful, unnecessary, and expensive surgical procedures.
How can unnecessary biopsies and surgical procedures be eliminated, all while improving early detection of cancers such as breast cancer?
Recent research might have the answer
Researchers from the Computer Science and Artificial Intelligence Laboratory (CSAIL) at MIT, Harvard Medical School, and Massachusetts General Hospital, believe that the answer lies in the use of artificial intelligence (AI). In order to improve both detection and diagnosis, the researchers collaborated to develop a machine learning AI system that would be able to predict if high-risk lesions detected by a needle biopsy following a mammogram would result in cancer.
Out of 334 high-risk lesions that were tested, the AI system correctly diagnosed 97% of breast cancers as malignant and thereby reduced the number of unnecessary surgeries by over 30%. “Because diagnostic tools are so inexact, there is an understandable tendency for doctors to over-screen for breast cancer. When there’s this much uncertainty in data, machine learning is exactly the tool that we need to improve detection and prevent over-treatment,” said Regina Barzilay, Delta Electronics Professor of Electrical Engineering and Computer Science at MIT.
The AI system is packed with information on over 600 existing high-risk lesions and it looks for patterns among the data available, including family history, demographics, past biopsies, and pathology reports. “To our knowledge, this is the first study to apply machine learning to the task of distinguishing high-risk lesions that need surgery from those that don’t. We believe this could support women to make more informed decisions about their treatment, and that we could provide more targeted approaches to health care in general,” said Constance Lehman, a professor at Harvard Medical School and chief of the Breast Imaging Division at MGH’s Department of Radiology.
Why false positives occur
When a mammogram detects a suspicious lesion, a needle biopsy needs to be performed in order to determine whether or not the lesion is cancerous. Although statistics show that over 70% of lesions are benign, 20% are malignant and 10% are high-risk lesions. Still, not all high-risk lesions need to be treated by means of surgery. Some doctors perform surgery on lesions that have increased cancer rates, such as a lobular carcinoma in situ (LCIS) or atypical ductal hyperplasia (ADH). “The vast majority of patients with high-risk lesions do not have cancer, and we’re trying to find the few that do. In a scenario like this there’s always a risk that when you try to increase the number of cancers you can identify, you’ll also increase the number of false positives you find,” said Manisha Bahl, co-author of the study and fellow doctor at MGH’s Department of Radiology.
Read on to learn more.