- Authors: Anh Phong Tran <tran.anh at husky.neu.edu> and Qianqian Fang <q.fang at neu.edu>
- Computational Optics and Translational Imaging (COTI) Lab, http://fanglab.org
- Northeastern University
- 360 Huntington Ave, Boston, MA 02115
- Version: 0.6
- License: CC-BY-SA (Creative Commons-Attribution-ShareAlike 4.0)
- URL: http://mcx.space/brain2mesh
- Project website: http://neurojson.org
The NeuroJSON JData/JMesh specifications are now officially funded by the National Institute of Health (NIH) and National Institute of Neurological Disorders and Stroke (NINDS) under grant# U24-NS124027.
Starting from V0.6, we include the originally excluded scalp, skull and CSF mesh layers for all Neurodevelopmental MRI atlas meshes. The segmentations of the CSF and skull layers were created using SPM12. For atlases of younger populations, the SPM segmentations may not be accurate as they were computed based on an adult head template. The scalp surfaces were extracted from the T1 MRI image using a threshold-based method. Details regarding processing steps of these tissue types can be found in Section 3.3 and Fig. 9 of our paper.
Please be cautious when using meshes of these tissue layers, especially in the younger populations, in your analyses. Although they were generated using the best tools that are currently available, understanding their known limitations helps to avoid inaccurate conclusions derived from this mesh data library.
This library is described in the following paper:
- Anh Phong Tran†, Shijie Yan†, Qianqian Fang*, (2020) "Improving model-based fNIRS analysis using mesh-based anatomical and light-transport models," Neurophotonics, 7(1), 015008, URL: http://dx.doi.org/10.1117/1.NPh.7.1.015008
To download the latest version of the entire brain mesh library, you may use the below link
https://github.com/NeuroJSON/BrainMeshLibrary/archive/master.zip
or use the below git command
git clone https://github.com/NeuroJSON/BrainMeshLibrary.git
To download the latest release (stable version), please browse
https://github.com/NeuroJSON/BrainMeshLibrary/releases
If you only need to download a particular brain mesh model, you may browse https://github.com/NeuroJSON/BrainMeshLibrary and select the mesh file (.jmsh) and click on the "Download" button.
The mesh data in this mesh library are stored in the JSON/JMesh format. The JMesh data format is a JSON-compatible data annotation open-standard defined in the JMesh Specification (http://github.com/NeuroJSON/jmesh).
The data files in this library are essentially JSON files. One can open these files using any JSON parser, widely avaialble for many programming languages. The data in these JSON files are zlib-compressed and base64-encoded. We have developed several open-source JMesh/JData decoders to load these data files conveniently.
To load the mesh files in MATLAB/GNU Octave, one need to install the JSONLab and ZMat toolboxes, from the below URLs:
- JSONLab: https://github.com/fangq/jsonlab/releases/latest
- ZMat: https://github.com/fangq/zmat/releases/latest
Once the toolboxes are downloaded and uncompressed, one need
to run addpath in MATLAB to add the path to the toolboxes,
then, the data files can be loaded by
>> data=loadjson('/path/to/filename.jmsh')
data =
x0x5F_DataInfo_: [1x1 struct]
MeshNode: [256537x3 single]
MeshElem: [1567340x5 uint32]
where the /path/to/filename.jmsh should be replaced by the
relative or full path to the .jmsh file from the library.
The loaded mesh contains a metadata record (x0x5F_DataInfo_
in MATLAB and _DataInfo_ in Octave), storing the mesh
information such as version and atlas name etc.
The data.MeshNode is an N x 3 matrix storing the
x/y/z coordinates of all the nodes in the mesh, and
data.MeshElem stores a 5-column integer array, with the
first 4 columns denoting the indices of the tetrahedral
elements and the last column denoting the region label.
The meanings of the labels are defined in the table
stored as data.x0x5F_DataInfo_.ROIBrainROILabels.
To plot or process the loaded mesh, one need to install the Iso2Mesh toolbox from
- Iso2Mesh: https://github.com/fangq/iso2mesh/
Similarly, after installing the toolbox, one need to run
plotmesh(data.MeshNode, data.MeshElem);
or
plotmesh(data.MeshNode, data.MeshElem,'x>100');
to plot a cross-section of the mesh.
To plot a particular tissue segment, one can use the 5th column in the MeshElem array, for example, to plot the white-matter (WM) segment, one can use
plotmesh(data.MeshNode, data.MeshElem(data.MeshElem(:,5)==5,:),'x>100');
This project is funded by the National Institutes of Health (NIH) / National Institute of Neurological Disorders and Stroke (NINDS) grant# U24-NS124027 and R01-EB026998 (PI: Fang), National Institute of General Medical Sciences (NIGMS) grant# R01-GM114365 (PI: Fang).