Nonlinear least squares fitting for myelin water imaging data acquired by mGRE sequence (3-pool model)
- Input data: multi-echo gradient echo (mGRE) data.
- Run preprocessing.m first and then run ifield_fit.m.
- Fitting output
- Complex model: amplitude, t2star, and frequency shift for each of the 3 water pool, and one overall initial phase.
- Magnitude model: amplitude and t2star for each of the 3 water pools.
- Initialization parameters of the fitting may need to be tuned accordingly for best performance.
- The 3-pool model is desribed in this paper: https://doi.org/10.1016/j.neuroimage.2015.03.081
- preprocessing.m: correct raw bipolar mGRE k-space data, produce voxel-wise complex data, fit total field map and R2* map. Necessary functions in this script can be found at https://github.com/hanwencat/QSM_Bruker
- ifield_fit.m: main fitting program, require preprocessed inputs: complex signal, total field map, and R2* map.
- objfun_complex_model.m: objective function for a 3-pool complex model (default in ifield_fit.m).
- objfun_magnitude_model.m: objective function for a 3-pool magnitude model.
- phantom_make: function to produce a 2D computational phantom.
- phantom_fit: script to test the fitting performance using the computational phantom.