dHCP_neonatal_HARDI_atlas

dHCP neonatal-multi-component-HARDI-atlas

This hosts the multi-component atlas of diffusion MRI data over the neonatal period created for the publication A framework for multi-component analysis of diffusion MRI data over the neonatal period. doi: 10.1016/j.neuroimage.2018.10.060 using data collected as part of the developing Human Connectome Project (dHCP).

The weekly templates can be downloaded individually from here.

---

To conveniently get local access to all data, I'd recommend using the GIN-CLI client.

On macOS, it can be installed via homebrew:

brew tap g-node/pkg
brew install g-node/pkg/gin-cli

On ubuntu:

wget https://web.gin.g-node.org/G-Node/gin-cli-releases/raw/master/gin-cli-latest.deb
sudo dpkg -i gin-cli-latest.deb
sudo apt install -f

For standlone binaries and instructions for other operating systems, see here.

Then, to download this repository simply execute

gin get maxpietsch/dHCP_neonatal_HARDI_atlas
cd dHCP_neonatal_HARDI_atlas

Initially, place-holder files are downloaded instead of larger files (.git/annex). If you want to download all files use gin download --content, for individual files or directories use gin get-content <location>, for downloading all rigidly aligned templates use gin get-content v1/rigid/mean, check the download status with gin ls. See GIN CLI Usage Tutorial for details.

This repo is organised hierarchically:

├── LICENSE
├── README.md
└── v1                          <-- atlas version (`v1` is as used for the paper)
    ├── joint                   <-- jointly aligned data (single space across time)
    │   ├── mean                <-- mode of aggregation
    │   ├── median              <-- mode of aggregation
    │   ├── rois                <-- regions of interest used for the paper
    ├── rigid                   <-- weekly templates, rigidly aligned
    │   ├── mean
    │   └── median
    └── metadata                <-- subject information and response functions
        ├── rf_csf_all
        ├── rf_native
        ├── rf_tissue_32.9
        ├── rf_tissue_44.1
        ├── sub_ses_ga_32.9
        ├── sub_ses_ga_34.0
        ├── sub_ses_ga_35.2
        ├── sub_ses_ga_35.7
        ├── sub_ses_ga_37.1
        ├── sub_ses_ga_38.1
        ├── sub_ses_ga_39.1
        ├── sub_ses_ga_40.1
        ├── sub_ses_ga_40.9
        ├── sub_ses_ga_42.0
        ├── sub_ses_ga_42.8
        └── sub_ses_ga_44.1

The files (see fig. 3):

├── LICENSE
├── README.md
└── v1
    ├── joint
    │   ├── mean
    │   │   ├── 32.9
    │   │   │   ├── adc.mif.gz                         <-- DTI-fit, ADC map
    │   │   │   ├── count.mif.gz                       <-- sum across masks, useful to modify template mask
    │   │   │   ├── fa.mif.gz                          <-- DTI-fit, FA map
    │   │   │   ├── mask.mif.gz
    │   │   │   ├── odf_csf_normed_cyo.mif.gz          <-- fluid component ODF, intensity, normalised
    │   │   │   ├── odf_csf_normed_native.mif.gz       <-- subject-native fluid image
    │   │   │   ├── odf_o_normed_cyo.mif.gz            <-- Ao: older-appearing tissue component
    │   │   │   ├── odf_wm_gm_normed_native.mif.gz     <-- subject-native tissue image
    │   │   │   └── odf_y_normed_cyo.mif.gz            <-- Ay: younger-appearing tissue component, normalised
    │   │   ├── 34.0
...
    ├── metadata
    │   ├── rf_native                                   <-- subject-specific response functions
    │   │   ├── sub-CC00063AN06_ses-15102_36_csf.txt    <-- CSF response function for this subject (PMA: 36w)
    │   │   ├── sub-CC00063AN06_ses-15102_36_wm_gmi.txt
    │   │   ├── sub-CC00065XX08_ses-18600_41_csf.txt
    │   │   ├── sub-CC00065XX08_ses-18600_41_wm_gmi.txt
    │   │   ├── sub-CC00067XX10_ses-20200_40_csf.txt

...
    │   ├── rf_csf_all                                  <-- group-average response functions
    │   ├── rf_tissue_32.9
    │   ├── rf_tissue_44.1
    │   ├── sub_ses_ga_32.9                             <-- subject identitifers per age group
    │   ├── sub_ses_ga_34.0
    │   ├── sub_ses_ga_35.2
    │   ├── sub_ses_ga_35.7
    │   ├── sub_ses_ga_37.1
    │   ├── sub_ses_ga_38.1
    │   ├── sub_ses_ga_39.1
    │   ├── sub_ses_ga_40.1
    │   ├── sub_ses_ga_40.9
    │   ├── sub_ses_ga_42.0
    │   ├── sub_ses_ga_42.8
    │   └── sub_ses_ga_44.1
    └── rigid
        ├── mean
        │   ├── 32.9
        │   │   ├── adc.mif.gz
        │   │   ├── count.mif.gz
        │   │   ├── fa.mif.gz
        │   │   ├── mask.mif.gz
        │   │   ├── odf_csf_normed_cyo.mif.gz
        │   │   ├── odf_csf_normed_native.mif.gz
        │   │   ├── odf_o_normed_cyo.mif.gz
        │   │   ├── odf_wm_gm_normed_native.mif.gz
        │   │   └── odf_y_normed_cyo.mif.gz
        │   ├── 34.0
...

Note that mean and median aggregation was perfomed using (Not-a-Number-) masked data using the brain masks as inclusion criteria for the aggregation operation. This prevents bias due to cropped data and allows retrospectively thresholding the template voxels by the number of subjects using the count.mif.gz images. For instance, the following replaces voxels to which fewer than 5 subjects contribute by NaNs: mrcalc count.mif.gz 5 -ge odf_o_normed_cyo.mif.gz nan -if odf_o_normed_cyo_thresholded.mif.gz

You can cite this work:

Pietsch, Maximilian, et al. "A framework for multi-component analysis of diffusion MRI data over the neonatal period." NeuroImage 186 (2019): 321-337.

@article{pietsch2019framework,
  title={A framework for multi-component analysis of diffusion MRI data over the neonatal period},
  author={Pietsch, Maximilian and Christiaens, Daan and Hutter, Jana and Cordero-Grande, Lucilio and Price, Anthony N and Hughes, Emer and Edwards, A David and Hajnal, Joseph V and Counsell, Serena J and Tournier, J-Donald},
  journal={NeuroImage},
  volume={186},
  pages={321--337},
  year={2019},
  publisher={Elsevier}
}