Written by IEEE | October 23, 2017
October is Breast Cancer Awareness Month, and, as it has with other diseases, technology has been responsible for advancements in early detection and treatment. While women, regardless of geographic or socioeconomic situation, are at risk for developing breast cancer, men also have a lifetime risk of 1 in 1,000. Almost 50 percent of breast cancer cases and 58 percent of deaths occur in less-developed countries.
In one such study, researchers at MIT have been using AI to predict whether or not breast lesions from a biopsy will be cancerous. High-risk lesions are abnormal cells found in a breast biopsy and appear suspicious, but many times they turn out to not be dangerous. Historically, this is only determined after an operation.
Using AI and machine learning could help to distinguish which lesions could become breast cancer and need to be removed, as well as which lesions are benign, saving people from unnecessary surgery. A machine learning system has been created to learn information about lesions, as well as demographics, family history, biopsies, type of lesion and pathology reports, and look for patterns that can better predict the chances of a lesion developing into breast cancer.
The first test of this study was on 335 lesions. The system accurately predicted 37 of 38 lesions that developed into cancer. This shows promising progress in the fight against cancer and increased quality of life for those undergoing treatment.