Real-time AI integration for MR to detect artifacts and guide pulse sequence adaptations
- Open access
AI-integrated MR pulse sequence reduces out-of-voxel artifacts by dynamically updating gradient schemes within 2 seconds, improving spectral quality.
- Why it matters: Out-of-voxel artifacts compromise magnetic resonance spectroscopy accuracy, and current methods lack real-time detection and correction, limiting clinical and research applications.
- What they did: The team developed PEREGRINE, a system combining convolutional autoencoders and neural networks to detect OOV artifacts in real-time during MR scans, testing 48 gradient permutations across multiple transients.
- The result: AI-driven updates significantly lowered OOV scores during persistent artifact conditions and improved fit quality, enabling more reliable spectral data and advancing real-time artifact management in MR imaging.