Together with Sage Bionetworks, Kaiser Permanente Washington Health Research Institute, and the University of Washington School of Medicine, IBM Research has unveiled a combination of machine learning algorithms and assessments by radiologists, in order to improve the accuracy of breast cancer screenings.
Radiologists often use Mammogram screenings in order to detect signs of breast cancer. The screenings help in visually identifying the potential hints of cancer.
“Through the current state of human interpretation of mammography images, two things happen: Misdiagnosis in terms of missing the cancer and also diagnosing cancer when it’s not there,” explained IBM researcher Stefan Harrer.
“Both cases are highly undesirable — you never want to miss a cancer when it’s there, but also if you’re diagnosing a cancer and it’s not there, it creates enormous pressure on patients, on the healthcare system, that could be avoided.
“That is exactly where we aim to improve things through the incorporation of AI (artificial intelligence) to decrease the rate of false positives, which is the diagnosis of cancer, and also to decrease missing the cancer when there is one.”
For the latest study, the team used more than 310,800 de-identified mammograms and clinical data from Kaiser Permanente Washington (KPWA) and the Karolinska Institute (KI) in Sweden. Out of the combined datasets, KI supplied around 166,500 examinations from 6,800 women, from which 780 were diagnosed positive; whereas the remaining 144,200 examinations were provided by KPWA from 85,500 women, of which 941 were cancer positive.
According to Harrer, application of AI in interpreting mammograms was not a recent venture, however the latest study was significant, mainly because of the size of cases it studied.
Furthermore, owning to the latest research, the team created an ecosystem which provides access to datasets which were previously unavailable for research activities.
“What we’ve done is create an ecosystem that allows us to keep that dataset … behind a secure firewall … to allow researchers to build models and submit these models to us, as the organisers of this ecosystem,” Harrer explained.
“These models can then come through the data and be tested, trained, and validated inside this secure environment by us, and then the performance of these models be returned back to the researchers and they can keep on running and improving the model.”