Image reconstruction using a gradient impulse response model for trajectory prediction.

Area of Science:

• Magnetic Resonance Imaging (MRI)
• Medical Physics
• Image Reconstruction

Background:

• Gradient imperfections are a significant challenge in MRI, particularly affecting sequences with long data acquisition times.
• Accurate k-space trajectory prediction is crucial for mitigating artifacts and improving image fidelity.

Purpose of the Study:

• To investigate image reconstruction methods utilizing k-space trajectories predicted by a gradient system's impulse response model.
• To assess the effectiveness of this model-based approach in improving image quality for challenging MRI sequences.

Main Methods:

• Gradient system characterization was performed, and impulse response functions were used to predict k-space trajectories for single-shot EPI, spiral, and variable-speed EPI sequences.
• Image reconstruction was conducted using predicted trajectories on phantom and in vivo data, with comparisons to concurrent field monitoring, separate trajectory measurements, and nominal trajectories.

Main Results:

• Model-based trajectory prediction resulted in high-quality MR images, comparable to reconstructions using dedicated trajectory measurements.
• This approach substantially reduced ghosting, blurring, and geometric distortion compared to using nominal trajectories.
• Image quality remained consistent, whether using recent or 3-year-old gradient system characterization data.

Conclusions:

• Model-based trajectory prediction facilitates high-quality image reconstruction for demanding MRI sequences like single-shot EPI and spiral imaging.
• This technique shows significant potential for accelerating structural imaging and advancing neuroimaging applications such as fMRI, DTI, and ASL.
• The method's reliance on a one-time system characterization, even with historical data, enhances its practical applicability.