Diffusion-weighted imaging processing

This section describes all DWI-related pre-processing steps implemented in micapipe, which heavily relies on tools from mrtrix. This includes image processing in preparation for the construction of tractography-based structural connectivity matrices, as well as associated edge length matrices. This processing pipeline has been optimized for multi-shell DWI, but can also handle single-shell data.



This module performs required pre-processing of DWI scans, in addition to deriving useful metrics from diffusion images (e.g. fractional anisotropy, mean diffusivity). For multi-shell data, the Dhollander algorithm is applied to estimate response functions of CSF, gray, and white matter, and multi-shell, multi-tissue constrained spherical deconvolution is used to estimate fibre orientation distributions. For single-shell data, the Tournier algorithm and single-shell, single tissue constrained spherical deconvolution are used for these processing steps, respectively.

Prerequisites 🖐🏼

You need to run -proc_structural before this stage

  • All DWI scans found in the bids directory are aligned using a rigid-body registration, and concatenated.

  • Concatenated DWI images undergo denoising by MP-PCA and Gibbs ring correction. Residuals are also calculated from denoised images

  • Correction of eddy current-induced distortions and motion

  • Non uniformity bias field correction

  • b0 image is linearly registered to the structural image (nativepro)

  • DWI brain mask is generated by registering MNI152 brain mask to DWI space using previously generated transformations

  • Compute fractional anisotropy and mean diffusivity images

  • Estimate response functions of CSF, gray, and white matter for spherical deconvolution

  • Estimate fibre orientation distributions using spherical deconvolution

  • Compute non-linear transformation from DWI (using white matter fibre orientation distribution image) and structural image aligned to the b0 scan

  • Apply inverse non-linear transformation to 5-tissue-type images for anatomically-constrained tractography

  • Compute track density image with 1 million streamlines using the iFOD1 algorithm. This image is mainly generated for quality control of previous DWI pre-processing.



This modules computes tractography-based structural connectivity matrices and associated edge length matrices. We apply iFOD2 for this purpose, a probabilistic tractography algorithm.

Prerequisites 🖐🏼

You need to run -proc_structural, -proc_surf, -post_structural, and -proc_dwi before this stage

  • Compute tractogram with specified number of streamlines using iFOD2 algorithm

  • Build structural connectomes and edge length matrices from cortical, subcortical, and cerebellar parcellations non-linearly registered to DWI space

  • If requested, compute automatic bundle segmentation using auto tractography.