RSNA names winners of intracranial hemorrhage AI challenge
The Radiological Society of North America (RSNA) has named the top 10 finishers of its most recent AI competition.
The RSNA Intracranial Hemorrhage Detection and Classification Challenge, which kicked off in September, tasked teams with developing an algorithm capable of identifying and classifying subtypes of hemorrhages on head CT images. To do so, they were given a dataset of more than 25,000 scans.
Teams completed the challenge on a platform created by Kaggle—a subsidiary of Alphabet Inc.—which allowed access to the dataset and a virtual discussion forum. Kaggle also awarded a $25,000 prize that will be shared among the winning entries.
“The results produced by the winning teams achieved really impressive performance,” Luciano M. Prevedello, MD, chair of the Machine Learning Steering Subcommittee of the Radiology Informatics Committee (RIC), said in a statement. “The challenge demonstrates the increasing sophistication of the imaging AI research community and the real potential of this technology to improve the efficiency and quality of care in radiology.”
The top teams were announced Monday, Dec. 2, at the 105th RSNA Scientific Assembly and Annual Meeting, in Chicago. They are:
- SeuTao
- NoBrainer
- takuoko
- GZ
- Keep Digging Gold
- BRAINSCAN.AI
- Big Head
- 賞金で焼肉
- Mind Blowers
- VinBDI.MedicalImagingTeam
In order to provide an accurate dataset, volunteer specialists worked with the Machine Learning Steering Subcommittee and the Machine Learning Data Standards Subcommittee to label the more than 25,000 CT scans.
Charles E. Kahn Jr., MD, chair of the RSNA RIC and editor of the journal Radiology: Artificial Intelligence, called the effort “truly remarkable,” and went on to describe the dataset as a “resource of tremendous value for imaging research.”