Simplex Mesh Signal Representation for Modeling Susceptibility-Induced Distortion

Motivation: To correct image distortion arising from susceptibility-induced field inhomogeneity in spin-echo sequences. Goal(s): To develop a continuous signal representation to efficiently model the impact of static local field gradients on spatial encoding. Approach: A simplex mesh signal representation is developed to model the imaging subject and background field as a continuous piecewise linear function. The model is compared to a conventional discrete spins representation for forward simulation and image estimation tasks. Results are demonstrated with a simulated spin-echo sequence near metal. Results: The proposed method accurately models susceptibility-induced distortion with fewer parameters/pixel than the conventional approach for simulation and estimation tasks. Impact: A continuous piecewise linear signal representation (simplex mesh) improves forward modeling efficiency in the presence of background field gradients. This work can help improve multi-readout distortion correction methods for FSE near metal and EPI in the brain.