Christoph Lee
University of Wisconsin-Madison, WI·Medicine & Clinical Research
Screening mammograms save lives, but they are imperfect. In the U.S.
- Fields
Medicine & Clinical Research
Public Health & Epidemiology
Computer Science
- Topics
breast cancer screening
mammography
digital breast tomosynthesis
- Grants
- 3 known grants, through Jun 2027
What they work on
- Validate a 2D mammography AI algorithm from the DREAM Digital Mammography Challenge on UCLA's Athena Breast Health Network registry, and refine it with radiologist supervision plus genetic mutation and tumor subtype data.
- Scale the AI algorithm from 2D to 3D mammography, and run reader studies with academic and community radiologists to find the best way to present AI outputs during interpretation.
- Independently validate four commercial AI tools for 3D digital breast tomosynthesis (DBT) screening using a model-to-data approach on data from seven U.S. breast imaging registries, comparing AI with radiologists across patient and tumor characteristics.
- Use a validated breast cancer microsimulation model to estimate the long-term health benefits, harms and costs of AI-driven DBT screening, as a standalone reader or as a second reader, and test AI triage of likely-negative exams.
- Identify subpopulations of women with lower access to diagnostic breast imaging, and differences in diagnostic outcomes and follow-up timeliness, using Breast Cancer Surveillance Consortium data and multi-level and mediation statistical models.
More about the research
each year, about 30,000 breast cancers are missed and about 3.8 million screening exams wrongly suggest cancer. Many women who do get an abnormal result also face delays or gaps in follow-up care. The lab's projects ask how artificial intelligence (AI) and better care delivery can make breast cancer screening and diagnosis more accurate and fair. The work moves from building and testing AI tools to measuring their population-level effects and studying the care that follows an abnormal mammogram.
Why it matters
About 40 million U.S. women are screened for breast cancer each year, so even small accuracy gains could mean fewer missed cancers and fewer false alarms. Independent evidence on commercial AI tools can guide women, physicians, payers and policymakers before these tools are widely adopted. New quality-of-care indicators for diagnostic follow-up could support practice-level fixes and national guidelines. These could help close gaps for rural, lower-income and underinsured women.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Population-Based Evaluation of Artificial Intelligence for Mammography Priorto Widespread Clinical Translation
NIH · Jul 2022 – Jun 2027
Improving Outcomes in the Breast Cancer Screening and Diagnostic Continuum
NIH · Jun 2025 – May 2027
Artificial Intelligence for Improved Breast Cancer Screening Accuracy: External Validation, Refinement, and Clinical Translation
NIH · Mar 2026 – Dec 2026
Closest labs
Karla Kerlikowske
University of California-Davis, CA·Medicine & Clinical Research
Develop and validate models that predict a woman's 6-year risk of advanced breast cancer, combining clinical risk factors, race and ethnicity, imaging features and AI output in Breast Cancer Surveillance Consortium data, and test their fairness across racial and ethnic groups.
Joann Elmore
University of California-Los Angeles, CA·Medicine & Clinical Research
Run a two-year pragmatic randomized trial of more than 154,000 3D screening mammograms assigned to interpretation with or without an FDA-cleared AI tool, measuring cancer detection and recall rates across 16 breast imaging facilities.
Wei Zhao
Stony Brook University, NY·Biomedical Engineering
Design, optimize and fabricate a prototype dual-layer flat-panel detector that records two mammograms at different x-ray energies in a single exposure, giving perfectly aligned images free of patient-motion artifacts.
The school
University of Wisconsin-Madison: 91 labs on PI Pages, $1.9B a year on research, 854 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.