Simulation of diffusion MRI sequences and PSF analysis

This MATLAB code simulates dMRI sequences with echo planar imaging (EPI) and spiral trajectories and characterizes the resulting point spread function (PSF).

Requirements

• The spiral trajectories are generated using Brian Hargreaves’ code, which can be found here: mrsrl.stanford.edu/~brian/vdspiral/

• The 2D PSF analysis and image reconstruction requires a GPU.

• The images are reconstructed using the expanded signal model, which can be found here: github.com/TardifLab/ESM_image_reconstruction

PSF analysis

Effective resolution

The effective resolution of single-shot readout trajectories is measured using the Full Width at Half Maximum (FWHM) in phase-encode direction (PE). Follow steps in example_FWHM.m to learn how to use the code for FWHM calculation.

Specificity and sharpening factor

The accuracy of spatial encoding is further characterized by measuring the specificity and sharpening factor of the 2D PSF. Follow example_PSF_2D.m for the definitions of specificity and sharpening factor.

Phantom simulation

A digital brain phantom and coil sensitivity maps are available to simulate the image reconstruction pipeline and evaluate its impact on image quality. See example_phantom_sim.m for an example on how to use the code.

SNR analysis

Implementation of the multiple replica method (Robson et al., 2008) to measure SNR of MRI scans using raw k-space data and noise measurements. example_multiple_replice_recon.m includes the steps to reconstruct the multiple replicas.

Reference: Feizollah, S., Tardif, C. L. (2022). High-resolution diffusion-weighted imaging at 7 Tesla: single-shot readout trajectories and their impact on signal-to-noise ratio, spatial resolution and accuracy.

Link to the dataset: https://doi.org/10.5683/SP3/V7ITEH

For any question about the code please contact: sajjad.feizollah@mail.mcgill.ca