A team of researchers from the Massachusetts Institute of Technology MIT found a deep learning artificial intelligence platform had a high success rate in detecting breast cancer risk.
WHY IT MATTERS
The model retrospectively identified women at high risk of developing breast cancer from nearly 90,000 prior consecutive screening mammograms taken at Massachusetts General Hospital (MGH).
The deep learning model was able to correctly place 31 percent of all the patients who subsequently developed breast cancer, compared with just 18 percent achieved with the existing Tyrer-Cuzick model.
The study indicates a mammography-based deep learning model could provide more accurate risk prediction by identifying patterns indicative of breast cancer.
“We hypothesize that there are subtle but informative cues on mammograms that may not be discernible by humans or simple volume-of-density measurements, and deep learning (DL) can leverage these cues to yield improved risk models,” the published study explained.
Although mammographic density improves the accuracy of breast cancer risk models, the use of breast density is limited by subjective assessment, variation across radiologists, and restricted data.
Rather than manually identifying discriminative image patterns, the team relied on their machine learning model to discover these patterns directly from the data.
The research team’s mammography-based deep learning model was able to provide more accurate risk prediction, and their hybrid DL model is equally accurate for white and African American women — another improvement over the Tyrer-Cuzick model.
THE BIGGER TREND
“Since the 1960s radiologists have noticed that women have unique and widely variable patterns of breast tissue visible on the mammogram,” Constance Lehman, professor of radiology at Harvard Medical School, told News-Medical. “These patterns can represent the influence of genetics, hormones, pregnancy, lactation, diet, weight loss, and weight gain. We can now leverage this detailed information to be more precise in our risk assessment at the individual level.”
The study included women who had either a diagnosis of breast cancer within 5 years, or imaging follow-up for at least 5 years from the date of index mammography.
Risk factors used in the hybrid deep learning model included age, weight, height, menopausal status, a detailed family history of breast and ovarian cancer, BRCA mutation status and breast density, among other factors.
An editorial published in Radiology argued in support of the research with the position that in assessing cancer risk from mammograms deep learning is superior to conventional risk models.
An April report from the Radiological Society of North America (RSNA), which publishes the journal, indicated AI and machine learning technologies could improve medical imaging in several respects as well.
The study noted AI systems are already being developed to improve medical image reconstruction, reduce noise and provide quality assurance, as well as with computer-aided detection and computer-aided classification and radiogenomics.
Each of these technologies could be used to enhance data sets, data engineering, and data science that would lead to the successful deployment of AI applications in medical imaging.
Nathan Eddy is a healthcare and technology freelancer based in Berlin.
Email the writer: firstname.lastname@example.org
Healthcare IT News is a HIMSS Media publication.