An internationally recognized expert on breast magnetic resonance imaging (MRI), Linda Moy, MD, professor in the Department of Radiology and a member of NYU Langone Health’s Perlmutter Cancer Center, wears a number of hats. At the forefront of developing new technologies to help doctors detect and treat cancer early, she also serves as a senior deputy editor at the journal Radiology and is among the leaders at the American College of Radiology (ACR), which establishes breast cancer screening guidelines.

Radiologists currently use computer-aided detection (CAD), first-generation computer software that analyzes scans, such as mammograms or chest X-rays, and marks areas where there’s a probability of an abnormality. The radiologist then decides whether the prompt is a normal variant or something questionable that should be worked up further in a patient. CAD is helpful in detecting some tumors, but it also results in missed radiological diagnoses known as callbacks and false positives. Artificial intelligence (AI), Dr. Moy says, has the potential to detect cancer with fewer false positives.

To this end, Dr. Moy collaborates with researchers at NYU’s Center for Data Science to develop AI algorithms with the aim of helping radiologists improve breast cancer detection and limit the need for patients to return for additional imaging when results are not clear. The collaboration also focuses on developing models to predict disease prognosis and likelihood of recurrence in people who have cancer as well as improving estimates of a woman's lifetime risk of breast cancer.

Research using the system developed by Dr. Moy and her colleagues found that the AI tool can be trained to detect changes in breast cells that are invisible to the human eye. The study also found that when the AI tool was paired with a human radiologist, results were more accurate than either the AI tool or a human radiologist alone. 

The U.S. Food and Drug Administration now allows these systems to provide a probability of malignancy, which is more helpful to radiologists. Dr. Moy says the Department of Radiology plans to study the AI system she and her colleagues have developed to better understand its value when treating people who have cancer.

“There are questions such as does AI actually help the radiologists when we are seeing the films on patients and is this something that the patients trust and want to use,” Dr. Moy says. “There have been many published studies where the results look good, but we want to make sure that in fact those results are reproducible once we start using AI in patients.”

New Screening Guidelines Offer Clarity

As a member of the Commission on Breast Imaging and the Screening Leadership Group for the ACR and chair of ACR’s Appropriateness Criteria in Breast Imaging, Dr. Moy has helped to refine and consolidate the guidelines for breast imaging from various societies as well as federal guidelines from the U.S. Preventive Services Task Force. 

There has been some controversy regarding the frequency of mammograms, Dr. Moy says, with some recommendations encouraging annual and others biannual examinations. The age at which to begin screenings also varies, with some guidelines recommending screening when a woman turns 40 and others advising that a woman wait until she is 50.

“Obviously, there have been mixed and confusing messages,” Dr. Moy says. “What ACR is saying is that in this era of advocating for patients, there should be an informed decision between the patient and referring physician.”

ACR’s new guidelines recommend annual screening for breast cancer beginning at age 40, continuing until at least age 70.

“Studies have shown that when we find cancer early, our patients' survival is better,” Dr. Moy says. “The bottom line is that we think screening and finding cancer early, when it’s most treatable, gives women the best chance of beating it.”