Neocortical and subcortial feature post-processing with z-brains relies heavily on micapipe,
an openly available, multimodal MRI pre-processing pipeline. The sections below outline the minimal pre-processing that should be perfomed with
micapipe prior to running z-brains, for each modality available in your dataset.
The Brain Imaging Data Structure (BIDS)
The Brain Imaging Data Structure (BIDS) is a widely used directory and file naming convention in brain imaging. To run micapipe, and subsequently z-brains, your raw data needs to be conform to BIDS standards.
Core to the z-brains architecture are the computation and resampling of neocortical surface segmentations via micapipe. This allows for smoothing and normalization of available features across participants.
Generation of these features rely on the structural processing of micapipe
Key outputs of these modules for z-brains are:
| Outputs | Description |
|---|---|
| Processed T1w image | A new space, nativepro, is created and used for all downstream processing and cross-modality registrations. This image is generated from the intensity non-uniformity-corrected, rescaled raw T1w image. If multiple T1w acquisitions are available, their are co-registered and average prior to intensity correction steps. |
| Neocortical surface reconstructions | Neocortical surfaces are initially generated by FreeSurfer/FastSurfer (called through micapipe), and resampled to symmetric, standard templates (fs-LR) aligned to the participant's nativepro space. |
| Neocortical thickness | Thickness estimates are computed by FreeSurfer/FastSurfer, and mapped to resampled fs-LR surface templates |
| Subcortical structure volume | Subcortical volumes are provided by FreeSurfer/FastSurfer (called through micapipe) |
| Subcortical structure segmentation | Subcortical structures are segmented using FSL FIRST, and the resulting parcellation is used to sample image intensities for relevant features. |
Generation of these features rely on the FLAIR processing of micapipe
Key outputs of this module for z-brains are:
| Outputs | Description |
|---|---|
| Processed FLAIR image | Pre-processed FLAIR image (bias correction, intensity clamping and rescaling, and normalization to gray/white matter interface values) co-registered to nativepro space. This image is used for intensity sampling in z-brains subcortical and hippocampal processing modules. |
| Processed FLAIR intensities mapped to neocortical surface | Neocortical surfaces are initially generated by FreeSurfer/FastSurfer (called through micapipe), and resampled to symmetric, standard templates (fs-LR) aligned to the participant's nativepro space. |
Generation of these features rely on the Microstructural profile covariance module of micapipe
Key outputs of this module for z-brains are:
| Outputs | Description |
|---|---|
| Processed T1 map volume | The qT1 image co-registered to nativepro space. This image is used for intensity sampling in z-brains subcortical and hippocampal processing modules. |
| T1 relaxation times mapped to neocortical surfaces | Neocortical surfaces are initially generated by FreeSurfer/FastSurfer (called through micapipe), and resampled to symmetric, standard templates (fs-LR) aligned to the participant's nativepro space. |
Generation of these features rely on the diffusion-weighted imaging module of micapipe
Key outputs of this module for z-brains are:
| Outputs | Description |
|---|---|
| Processed MD volume | The MD image co-registered to nativepro space. This image is used for intensity sampling in z-brains subcortical and hippocampal processing modules. |
| Processed FA volume | The FA image co-registered to nativepro space. This image is used for intensity sampling in z-brains subcortical and hippocampal processing modules. |
| MD and FA values mapped to neocortical surfaces | Neocortical surfaces are initially generated by FreeSurfer/FastSurfer (called through micapipe), and resampled to symmetric, standard templates (fs-LR) aligned to the participant's nativepro space. |