Prediction of motion induced magnetic fields

Prediction of motion induced magnetic fields In MRI, B0 field inhomogeneity maps are frequently employed to enhance image quality. These inhomogeneities are influenced by the location of the head within the static magnetic field. Subject head movements, which are prevalent in lengthy experiments like fMRI, can disrupt the inhomogeneity distribution. This is because head motion alters the relative positions and orientations of susceptibility interfaces within the static B0 field. Consequently, a single field map acquired at the beginning of an fMRI session may not be adequate for correcting geometric distortions throughout the entire experiment. Acquiring a whole-brain field map using dual-echo GRE typically takes 1 to 2 minutes, depending on the field-of-view (FOV), resolution, and acceleration provided by parallel imaging. Due to time limitations, repeating this procedure multiple times for each potential head position may not be feasible.

In this work, we introduce a method for simulating field maps to anticipate variations in the B0 field at various head positions. This approach takes into account both global and more local effects using a susceptibility model derived from ultra-short echo time (UTE) scans. The UTE sequence proves effective in detecting signals from tissues with extremely brief T2 components (e.g., bone) where TEs are less than 1 ms.

Our systematic approach encompasses four key components for predicting the field map:

1. A multi-class tissue model is employed to estimate fields induced by different types of tissues.
2. A linear k-space model is utilized to account for gradient imperfections.
3. Dipole estimation is employed to quantify perturbing fields from the lower body.
4. A position-dependent tissue mask is integrated to model alterations in the field map due to large motion effects.

When combined with a head tracking technique, our methodology has the potential to make real-time shim current calculations in practice.