MRNet

Introduced 2018-11-27

The MRNet dataset consists of 1,370 knee MRI exams performed at Stanford University Medical Center. The dataset contains 1,104 (80.6%) abnormal exams, with 319 (23.3%) ACL tears and 508 (37.1%) meniscal tears; labels were obtained through manual extraction from clinical reports.

Source: MRNet

Benchmarks

Multi-Label Classification/Average AUC Multi-Label Classification/AUC on Abnormality (ABN)Multi-Label Classification/AUC on ACL Tear (ACL)Multi-Label Classification/AUC on Meniscus Tear (MEN)Multi-Label Classification/Average Accuracy Multi-Label Classification/Accuracy on Abnormality (ABN)Multi-Label Classification/Accuracy on ACL Tear (ACL)Multi-Label Classification/Accuracy on Meniscus Tear (MEN)