Dataset
RadImageNet is a large database of annotated medical images from multiple modalities and of multiple pathologies. The data can be licensed for commercial use. Use the contact form to request access to the academic dataset (reduced resolution, restricted labeling).
Large Size
Professionally Labeled
No Restrictions
RIN is comprised of 5 million labeled images of medical pathology with DICOM tags on over 1 million PET, CT, Ultrasound and MRI studies from 500,000 patients.
The images have all been labeled by fellowship trained, clinically practicing US radiologists with appointments at New York Hospital/Cornell Medicine
The dataset has been validated as HIPAA compliant
Images

Label: MCL Tear
TR: 2853
TE: 53.1
FA: 111
Slice thickness: 3.5
Window width: 5409
Window Center: 1902
Patient age: 36
Patient sex: Male
ROI coordinates: (347,352) (412,352)
(473,212) (473,379)
​

Label: ACL Tear
TR: 3408
TE: 58.6
FA: 111
Slice thickness: 3.5
Window width: 3496
Window Center: 1530
Patient age: 36
Patient sex: Male
ROI coordinates: (219,401) (379,482)
(419,238) (403,362)
​