
What is Spark?
Spark is a multi-departmental initiative at Duke with the mission of advancing research and implementation of AI in medical imaging.
We focus on problems in medicine where AI can make a significant impact while investing in technical innovations in the fields of machine learning and computer vision. We plan and conduct our research in close partnership with physicians from different clinical departments. We work closely with Duke AI Health.
Our Research
Our research focuses development and application of modern machine learning methods in the context of medicine with focus on medical imaging. Our research includes majority of imaging modalities (MRI, CT, X-ray, and ultrasound), and covers many clinical applications. Examples include prediction of adverse surgical outcomes, the quantification of breast density in MRIs, measurements related to musculoskeletal disorders, and analysis of thyroid nodules in ultrasound.
Our methodological research is currently focused on building, evaluating, and fine-tuning vision foundation models, medical image segmentation, and medical image harmonization. We strongly believe in open science and you can find our models and data in the resources section.
A lot of our research is oriented around the The Human Body Project. We are funded by internal Duke sources and extramural grants.
Work with Duke Spark
Collaboration is at the core of the Duke Spark initiative. We are interested in collaborations with academia as well as industry partners. If your company is interested in the development or evaluation of AI algorithms in the context of medical imaging, contact us at maciej.mazurowski@duke.edu. Please also let us know if you are a faculty member or a student at Duke or an academic researcher at another institution interested in doing research with us.







