Brain-tumor AI
Automated tumor response assessment
Deep-learning tools that measure brain tumors on MRI and tell clinicians whether they are growing, shrinking, or stable.
A neuroradiology clinical research lab at Duke University
Building AI for brain imaging and better patient care.
We build artificial intelligence and computational image-processing tools that make brain MRI and CT more quantitative and more useful in the clinic. Our work focuses on brain tumors, pediatric and neonatal neuroimaging, and other CNS disease.
- Peer-reviewed papers
- 40+
- Public datasets & challenges
- 10+
- Trainees mentored
- 15+
- Core research themes
- 6
Our vision
Quantitative, trustworthy imaging for every patient
AI is changing radiology, but new tools only matter if they are accurate, fair, and usable in the clinic. We build automated image-analysis methods that give clinicians objective, reproducible measurements, from tracking whether a brain tumor is responding to treatment to predicting outcomes in newborns with brain injury. We also release the datasets and models we create so that other researchers can build on them.
Datasets & benchmarks
Public data that moves the field
We build and release large, expertly annotated imaging datasets and run international AI challenges (BraTS, FeTS).
From newborns up
Pediatric & neonatal neuroimaging
Predicting neurodevelopmental outcomes in newborns with brain injury and segmenting pediatric brain tumors.
