Inline AI on the MRI Scanner: from development to deployment
We develop AI for MRI and deploy it on clinical MRI scanners. Our AI runs inside the scanner, so the AI images appear on the scanner console during the exam. No data export or offline processing is needed, which is important for direct clinical use.
Inline AI
- 1DevelopTrain the AI on MRI data
- 2PackageAI model in a container
- 3DeployRun it on the scanner using Siemens FIRE3
- 4UseFacilitate clinical use of AI
Low resolution
CineGen, inline
Projects
1. CineGen: flow-matching generative high resolution cine MRI
CineGen is a conditional flow-matching super-resolution model for single-heartbeat free-breathing cine MRI. It runs inline on the scanner, in about one second per image.1,2
References
- Sun, C. Y.*, Wang, Y., Jiang, L., Grinstead, J., Chow, K., Bi, X., Baxter, S., Kumar, S., & Altes, T. A. (2026).
CineGen: Inline Conditional Flow-Matching Super-Resolution for Real-Time Cardiac Cine MRI.
ISMRM Workshop on Data Sampling and Image Reconstruction, Sedona, AZ, USA. Trailhead Award for Clinical Relevance and Impact. - Sun, C. Y.*, Wang, Y., Jiang, L., Grinstead, J., Chow, K., Bi, X., Baxter, S., Kumar, S., & Altes, T. A. (2026).
FIRE-integrated CineGen: inline conditional flow-matching super-resolution for real-time Cine MRI.
34th ISMRM, Cape Town, South Africa. Oral. - Chow, K., Kellman, P., & Xue, H. (2021).
Prototyping image reconstruction and analysis with FIRE.
SCMR 24th Annual Scientific Sessions, Virtual Meeting.