Longitudinal QSM (MATLAB)

MATLAB implementation of a longitudinal QSM reconstruction framework based on joint multi-time-point estimation, designed to improve inter-scan consistency while preserving sensitivity to true susceptibility changes.

Environment setup

This code is implemented in MATLAB.

The reconstruction is developed based on the MEDI QSM framework.
Please install the MEDI toolbox prior to running the code:

MEDI toolbox: https://pre.weill.cornell.edu/mri/pages/qsm.html

Make sure all required paths are properly added in MATLAB.

Usage

Run the main script to perform longitudinal QSM reconstruction of two time points.

Input

Required

• RDF1, RDF2
  Local field maps in radians
  (convert from Hz using 2π * delta_TE)

• Mask1, Mask2
  Brain masks

• voxel_size
  Voxel size (mm)

• delta_TE
  Echo spacing (s)

• CF
  Center frequency (Hz)

• B0_dir1, B0_dir2
  B0 direction vectors (unit vectors)

• iMag1, iMag2
  Magnitude images for MEDI regularization

• VesselMask
  Mask to exclude vessels from temporal sparsity

• Mask_CSF
  CSF mask for susceptibility referencing

• x0_init_ppm, dx_init_ppm
  Initial maps (ppm) (derived from MEDI)

All inputs are expected to be preprocessed (e.g., phase unwrapping, background field removal, and registration across time points).

License

© Jiye Kim
@ LIST, Seoul National University

This code is provided for academic and research purposes only.
Please cite or acknowledge the original author when using this work.