Using Artificial Intelligence to Identify Women for High-Risk Screening MRI
Description: The overarching goal of this Clinician Scientist Development Grant is to develop an Artificial Intelligence-based technology to better predict future risk of breast cancer based on mammography images and risk factors. Current American Cancer Society breast cancer screening guidelines recommend screening with magnetic resonance imaging (MRI) in women who have at least at 20% lifetime risk of breast cancer. However, current breast cancer risk models do not accurately predict future risk for individual women and perform worse in Black and Hispanic women than in White women. Artificial intelligence (AI)-based technologies are increasingly being used in health care to improve accuracy of medical testing and have shown promise as a risk prediction tool. Our goal is to develop an AI-based technology based on mammography images that predicts breast cancer risk accurately and equally across women of all racial/ethnic groups.
To accomplish this goal, first we will use our team of experts in breast imaging, machine learning, and technology assessment to develop the AI technology using a large dataset of mammography images from women receiving screening mammography at the University of Washington. Next, we will test the model using screening mammography images from the University of California at Los Angeles-Athena health network. We will compare the accuracy of the AI-based technology to other risk models currently used in clinical practice, and we will compare the accuracy across women of different racial and ethnic backgrounds. We will then use mathematical models to identify breast cancer risk thresholds that can be used to select women who may benefit from additional breast cancer screening with magnetic resonance imaging (MRI).
The results of this work will provide information that can be used to guide breast cancer screening recommendations tailored to women’s personal risk of breast cancer. This award will also support the training and career development of the Principal Investigator through coursework and mentorship in AI training and evaluation and statistical methods as she transitions to a career as an independent research scientist.
Funder: American Cancer Society
Amount: $583,200
Funding type: award
Start date: 2022-01-01
End date: 2025-12-31
Investigators: Kathryn Lowry, Dushyant Sahani, Helen Powell, Vikram Jandhyala, Carol Rhodes, Joann Elmore, Natasha Stout, Christoph Lee, Michelle Chan
Institutions: University of Washington School of Medicine, University of Washington, Harvard Pilgrim Health Care, Inc
Topic: AI in cancer detection
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Funded outputs count: 3
Funder award ID: CSDG-21-078-01-CPSH
OpenAlex ID: G8687408567