FESTiVE: Self-calibrated Spatial-temporal Gradient Characterization for Non-cartesian Imaging on High Performance Systems
eess.SP
Submitted: 2026-10-30
Updated: 2026-11-05
License: http://creativecommons.org/licenses/by/4.0/
The gist: Gradient imperfections disrupt the encoding benefits of long-readout imaging applications on high-performance systems.
Abstract
Gradient imperfections disrupt the encoding benefits of long-readout imaging applications on high-performance systems. Currently, external field monitoring hardware provides the only reliable solution for general spatial-temporal gradient characterization. We present FESTiVE, a data-driven alternative to higher-order field monitoring. FESTiVE leverages parallel imaging to both implicitly model and explicitly recover field imperfections. Using a fast k-t mapping calibration scan, FESTiVE formulates a time-varying GRAPPA problem which maps k-space data corrupted by a time-varying gradient imperfection against a clean reference. This mapping is represented compactly with an implicit neural representation, from which the spatial-temporal fields are recovered in image space using coil sensitivities. In phantom, FESTiVE produces comparable field measurements to a Skope field camera across multiple non-cartesian 2D acquisition strategies and gradient specifications. In vivo, FESTiVE achieves state-of-the-art image quality for a high-performance diffusion spiral protocol on the head-only MAGNUS 3T, where higher-order phase imperfections are significant. At 7T, on the head-only Siemens TerraX system, FESTiVE enables 300μm spiral imaging where conventional field camera-based approaches fall below probe dephasing limits. In conclusion, FESTiVE enables robust high-resolution non-cartesian imaging on high-performance gradients, characterizing gradient imperfections using parallel imaging rather than field monitoring hardware.
Sources
- Implicit Representation of GRAPPA Kernels for Fast MRI Reconstruction
- Adam: A Method for Stochastic Optimization
Related papers
- Open-H-RF: A Large-Scale Ultrasound Raw Channel Dataset Initiative
- Scalable Attention for 5G NR Channel Estimation
- Low-Altitude Wireless Networks: A Comprehensive Survey
- Sensing-Assisted OFDM with Index Modulation: A Waveform for Optical Wireless Integrated Communication and Localisation
- Iteration-Free Candidate Pruning and an Index-Domain VLSI Architecture for Sort-Free MIMO Box Decoding up to 1024-QAM
- A Foundation Model for Large-Scale Wireless Network Planning, Operation and Optimization