Researchers at University of Washington and University of California, Los Angeles, have developed an artificial intelligence system that could help pathologists read biopsies more accurately, and lead to better detection and diagnosis of breast cancer.
The artificial intelligence system outperformed doctors in differentiating ductal carcinoma in situ (pictured) from atypia, one of the more challenging diagnoses in breast cancer cases. Elmore Lab/UCLA
Doctors examine images of breast tissue biopsies to diagnose breast cancer. But the differences between cancerous and benign images can be difficult for the human eye to classify. This new algorithm helps interpret them - and it does so nearly as accurately or better than an experienced pathologist, depending on the task. The research team published its results Aug. 9 in the journal JAMA Network Open.
"This work concentrated on how to capture the characteristics of the different diagnostic classes by analyzing the pattern of the tissue classes surrounding the ducts in whole-slide images of breast biopsies,” said co-author Linda Shapiro , a professor in both the UW’s Paul G. Allen School of Computer Science & Engineering and the UW’s electrical and computer engineering department. "My doctoral student, Ezgi Mercan , invented a novel descriptor called the structure feature that was able to represent these patterns in a compact way for use in machine learning.”
In 2015, a study from the UW School of Medicine found that pathologists often disagree on the interpretation of breast biopsies , which are performed on millions of women each year. The study revealed that diagnostic errors occurred for about one out of every six women who had a noninvasive type of breast cancer called " ductal carcinoma in situ.” In addition, incorrect diagnoses were given in about half of the biopsy cases with abnormal cells that are associated with a higher risk for breast cancer - a condition called breast atypia.
"Medical images of breast biopsies contain a great deal of complex data, and interpreting them can be very subjective,” said co-author Dr. Joann Elmore , a professor of medicine at the David Geffen School of Medicine at UCLA, who was previously a professor of internal medicine at the UW School of Medicine. "Distinguishing breast atypia from ductal carcinoma in situ is important clinically, but very challenging for pathologists. Sometimes doctors do not even agree with their previous diagnosis when they are shown the same case a year later.”
The scientists reasoned that artificial intelligence could provide more accurate readings consistently. It uses a large dataset that makes it possible for the machine learning system to recognize patterns associated with cancer that are difficult for doctors to see. After studying the strategies that the pathologists used during breast biopsy interpretations, the team developed image analysis methods tailored to address these challenges.
The team fed 240 breast biopsy images into a computer, training it to recognize patterns associated with several types of breast lesions, ranging from noncancerous and atypia to ductal carcinoma in situ and invasive breast cancer. The correct diagnoses were determined by a consensus among three expert pathologists.
To test the system, the researchers compared its readings to independent diagnoses made by 87 practicing U.S. pathologists who interpreted the same cases. The algorithm came close to performing as well as human doctors in differentiating cancer from non-cancer. But it outperformed doctors when differentiating ductal carcinoma in situ from atypia, correctly diagnosing pre-invasive breast cancer biopsies about 89% of the time, compared to 70% for pathologists.
"These results are very encouraging,” Elmore said. "There is low accuracy among practicing pathologists in the U.S. when it comes to the diagnosis of atypia and ductal carcinoma in situ, and the computer-based automated approach shows great promise.”
The researchers are already working on training the system to diagnose skin cancer.
Ezgi Mercan , a researcher at Seattle Children’s Hospital who completed this research as a doctoral student in the Allen School, is the first author on this paper. Other authors are Sachin Mehta , a doctoral student in the UW’s electrical and computer engineering department; Dr. Jamen Bartlett at Southern Ohio Pathology Consultants; and Dr. Donald Weaver at the University of Vermont.