Machine Learning Improves Diagnostic Accuracy of Breast Cancer MRI, Study Shows

Machine learning can be used to improve the diagnostic accuracy of dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) for breast cancer, according to a study in this week's Science Translational Medicine. While DCE-MRI has a high sensitivity for detecting breast cancer, it often leads to unnecessary biopsies, particularly in women with intermediate or average risk for the disease. To help improve the overall accuracy of DCE-MRI, New York University researchers developed a deep learning system to predict the probability of  breast cancer in patients undergoing DCE-MRI and showed that it achieves diagnostic accuracy equivalent to breast imaging experts for predicting the probability of the breast cancer presence. The team also showed that combining the system's predictions with radiologist decisions further improved classification accuracy. The work, the scientists write, "creates a foundation for deployment and prospective analysis of DL-based models for breast MRI."