CONP Portal | Dataset


Hippocampal morphology and cytoarchitecture in the 3D BigBrain
Creators: Jordan DeKraker, Jonathan C. Lau, Ali R. Khan, Kayla Ferko, Stefan Köhler
Principal Investigator: Ali R. Khan
Contact: Ali R. Khan, Khan Computational Imaging Lab, Robarts Research Institute, University of Western Ontario alik@robarts.ca
Licenses: CC BY-4.0
Version: 1.0.0
Modalities: Histology
Formats: NIfTI STL
Size: 14.0 GB
No of Files: 134
No of Subjects: 1
Metadata file: DATS.json
Other Dates: Start Date: 2019-04-01 00:00:00
Description:
This project's preprint can be found here: https://www.biorxiv.org/content/10.1101/599571v1. The goal of this project is to characterize the complex mesoscale folding of the hippocampus as well as histological features which define the hippocampal subfields using the open source dataset 3D BigBrain.

Dataset README information

README.md

BIDS_40um/sourcedata/ contains the original left and right hippocampal block nifti images (ftp://bigbrain.loris.ca/BigBrainRelease.2015/2D_Final_Sections/Coronal/Png/Full_Resolution/), which is publically available under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International Public License (https://bigbrain.loris.ca/main.php?test_name=license). BIDS_downsampled/sourcedata contains 80um resampled blocks, on which manual labelling of hippocampal and surrounding structures was performed (BIDS_downsampled/manual_masks/).

Laplacian solutions were obtained over the domain of hippocampal grey matter according to our previous publication (https://www.ncbi.nlm.nih.gov/pubmed/29175494), code available at https://github.com/jordandekraker/HippUnfolding (BIDS_downsampled/HippUnfold). These solutions were obtained in the downsampled image space due to memory limits, and were then upsampled via the upsample.m script, which populated the corresponding directories in BIDS_40um.

In parallel, an equivolume laminar solution was also obtained over the domain of hippocampal grey matter using code available at https://github.com/nighres/nighres. The exact code used to obtain these solutions can be found in BIDSscripts/nighres.

All analyses performed in the current publication can be replicated using the scripts found in BIDSscripts. Many of these scripts depend on the outputs from other scripts. For example, allfeatures.m collects, preprocesses, and plots the 5 morphological features computed in MorphologyFeatures.m as well as the 10 laminar features computed in LaminarFeatures.m. For ease of use, .mat files are provided alongside each script containing all variables needed to run downstream analyses.

BIDSscripts/figures contains visualizations of all analyses, as well as supplementary analyses and some which were not included with our publication because no conclusive assertions could be made form them.

Download Using DataLad

CircleCI status

The following instructions require a basic understanding of UNIX/LINUX command lines. Future portal functionality may include downloads directly from the web browser. Dataset download is currently enabled through DataLad.

Note: The conp-dataset requires version >=0.12.5 of DataLad and version >=8.20200309 of git-annex.

To install DataLad on your system, please refer to the install section of the DataLad Handbook (installation via miniconda is recommended in order to obtain the latest version of DataLad).

1) Initiate the CONP dataset

To initiate the CONP dataset (conp-dataset), run the following command in the directory where you want CONP datasets to be installed:

datalad install https://github.com/CONP-PCNO/conp-dataset.git

2) Install the Khanlab/BigBrainHippoUnfold dataset

To install the dataset, go into the created conp-dataset directory and run datalad install on the dataset Khanlab/BigBrainHippoUnfold:

cd conp-dataset
datalad install projects/Khanlab/BigBrainHippoUnfold

3) Download the Khanlab/BigBrainHippoUnfold dataset

Now that the DataLad dataset has been installed, go into the dataset directory under projects/Khanlab/BigBrainHippoUnfold.

cd projects/Khanlab/BigBrainHippoUnfold

Note that files visible in the dataset are symlinks and will need to be downloaded manually using the datalad get command in the dataset directory:

datalad get <filepath>

Note, if you run datalad get * command, all the files present in the dataset directory will be downloaded.

For more information on how DataLad works, please visit the DataLad Handbook documentation .