The American College of Radiology (ACR) has released a new software platform designed to enable radiologists to develop their own artificial intelligence (AI) algorithms.
The ACR's Data Science Institute said the new platform, ACR AI-LAB, will be offered for free to radiologists to enable them to develop AI models "at their own facilities, using their own data, to meet their own clinical needs," according to a press release. The algorithms can be developed behind institutional firewalls, preserving data security.
The ACR believes that the AI-LAB platform "democratizes" artificial intelligence by supporting radiology departments in the direct participation in the development life cycle for AI algorithms. The framework is open and vendor neutral, and through it radiologists will be able to learn about AI, contribute AI datasets, share AI algorithms, evaluate and develop AI models, and combine models though transfer learning and model ensembles.
The AI-LAB framework is also designed to support future regulatory initiatives from the U.S. Food and Drug Administration (FDA) using real-world data to improve and monitor AI algorithms. The initiative is designed to address issues that have slowed AI development, such as the challenge of gathering large datasets of medical images for testing and training AI algorithms.
The ACR will demonstrate AI-LAB at its annual meeting May 18 to 22 in Washington, DC.
