fit-diffusion-model

Matlab code and atlas data for fitting diffusion models to diffusion-weighted MRI data.

Introduction

This repository contains Matlab code to fit various diffusion models to diffusion-weighted MRI data. Models include the conventional single tensor model, ball-and-sticks models, and various two-tensor models. Estimation methods include non-linear least squares or maximum-likelihood estimation assuming Rician noise. Furthermore, the repository also contains a model complexity atlas and an orientation atlas computed from 500 elderly subjects [ref], for respectively regularization and model selection purposes. Sample scripts and a sample dataset are provided to demonstrate how to register the atlas data to the subject space, and how to fit various diffusion models to the data.

Instructions

For registering the atlas data to the subject space:

1. Install FSL and DTI-TK
2. Run the Bash script Sample_code_coregistration_with_dtitk.sh.

For fitting diffusion models to the diffusion-weighted MRI data:

1. Download 'nifti-tools' from Jimmy Shen and 'estimation tools' from Dirk Poot, and put them in your Matlab path. For convenience, these dependencies have also been included in the directory matlab_code.
2. Run the Matlab script Sample_code_estimation_with_prior.m.

Questions

For further questions, you may refer to the paper [ref] or contact me at joorarkesteijn(at)gmail.com.