Skip to main content
Wolfson Institute of Population Health

AI in Mammography

Centre for Evaluation and Methods

Funding Body:  Cancer Research UK

Principal Investigator:  Dr. Adam Brentnall, Prof. Giovanni Montana,

Overview

Breast density

Breast density (also known as mammographic density) is the amount of white and bright regions seen on a mammogram. High breast density can make it harder for doctors to detect breast cancer on a screening mammogram and also increases the risk of developing breast cancer.

Measuring breast density can become a useful health tool for women, however, the way it is measured still needs to be improved, and it is not clear how best to use the information to aid early detection and prevention of the disease.

Part of this project is framed in this context. We aim at finding more reliable ways of measuring breast density, and at evaluating the risk of not seeing cancer during the screening because of breast density.

Artificial intelligence (AI) in healthcare

The main ingredient of our project is developing AI algorithms to improve the risk assessment of breast cancer. With the increasing availability of medical data, AI has become an evolving trend in the medical field. In preventive health care, the wealth of available data provides opportunities for more accurate health monitoring. The social impact of using AI in healthcare is a complex matter that attracted the attention of governments, media, and society in general.

The project

Our project will develop an artificial intelligence system to design a risk-adapted screening programme, by assessing risk of breast cancer in a mammogram at screening, and the long-term risk of breast cancer following a negative screen by mammography.
The three questions we hope to answer are:

  • Can we improve the way we detect cancer in a mammogram at screening? How do we assess the risk?
  • Some women have very dense breasts, and this makes it difficult to spot the signs of breast cancer in the mammogram. A ``false negative'' is when a women is assessed as being at low risk for breast cancer, when in fact she is at high risk. How can we tell when a woman might be at risk of getting a false negative during a standard mammogram, and should be offered an alternative screening method?
  • What is the long-term risk of breast cancer following a negative screen by mammography? Who should we screen for breast cancer, and how often should screenings take place?

Further Information

Patient and Public Involvement group

The ultimate user of medical AI are patients. Thus the healthy development of AI in healthcare depends, to a certain extent, on researchers being able to understand the public’s views and communicate transparently about their research. This is why it is important to us to discuss with a PPI group about the best ways to communicate about complex artificial intelligence algorithms, and about the inclusion of a component based on artificial intelligence in long-term risk of breast cancer. 

Back to top