The Application of Machine Learning and CADe to Identify Women at Risk of Cardiovascular Disease from Breast-Screening Mammography

Maguire, Dominic

Abstract

Cardiovascular disease (CVD) is the leading cause of premature death in the United Kingdom, killing more than twice as many women as breast cancer. Conventional CVD risk factors have been shown to have less accuracy for females who are considered low-risk. Recently, researchers have noted that breast arterial calcification (BAC), which is regularly observed as an incidental finding on breast screening mammograms, could be used as a potential indicator of increased risk of developing CVD.

This project has developed three models using deep learning, a type of artificial intelligence, that can automatically identify the presence of BAC on a mammogram (radiological image of the breast), where it is situated, and how much is present. An anonymized mammogram dataset was used during the training of each model. Images were marked as having BAC or not and were validated by consultant radiologists. Manual tracing of areas of BAC on each image was also carried out under the guidance of a radiologist. Models were trained on a computer graphics card.

One model reported 80% accuracy for the presence of BAC. This was achieved by tuning several parameters such as the size and number of images used, the model’s shape and size, the length of time the model was trained and how quickly it learned. This approach was also applied to the BAC location and quantity models. The resulting models will allow the extension of breast screening to include vascular screening, identifying those women attending for breast screening who are at risk of CVD.

• Online Publication Date: Jun 30, 2022
• Publication Date: Jun 30, 2022
• Deposit Date: Jan 23, 2025
• DOI: https://doi.org/10.17866/rd.salford.20161376.v1
• Publisher URL: https://salford.figshare.com/articles/poster/The_Application_of_Machine_Learning_and_CADe_to_Identify_Women_at_Risk_of_Cardiovascular_Disease_from_Breast-Screening_Mammography/20161376
• Collection Date: Jun 30, 2022