Notes:
Brain Biomechanics Data:
Each folder corresponds to a single data set. The name of each
folder consists of the subject ID and a description of the motion
investigated: "rot" is neck rotation and "drop" is neck extension.
Each folder contains a number of files described below:
• N1234*T1d.nii is the defaced T1-weighted MPRAGE in the same space
as the displacement and strain data.
• N1234*seg.nii is the brain segmentation of the MPRAGE. The labels
are: 5=cerebrospinal fluid, 6=ventricles, 14=cerebellar gray
matter, 15=cerebral gray matter, 16=caudate, 17=thalamus,
18=putamen, 22=brainstem, 24=cerebellar white matter, 25=cerebral
white matter.
• N1234*FA.nii is the fractional anisotropy from the diffusion
tensor imaging data.
• N1234*DTI_EV.nii is the principal eigenvector of the diffusion
tensor.
• N1234*u/v/w*nii are the displacement components over time along
the x-, y-, and z-axis, respectively.
• Lagrangian displacement and strain components over time:
◦ u/v/w are the displacement with respect to the skull along the
x-, y-, and z-axes, respectively
◦ E1/E2/E3 are the principal strain components along vectors
V1/V2/V3, respectively
◦ Eoct is the octahedral shear strain
◦ Exx/Exy/Exz/Eyy/Eyz/Ezz are the Cartesian strain components
• N1234*PosVelAcc.mat is a Matlab file that contains four
variables: t (time), p (position), v (velocity), and a
(acceleration) that describe the average angular position,
velocity, and acceleration over time for the head cradle of the
device.
All of the data is in the same space. The T1 and segmentation are
3D volumes (x,y,z); the displacement and strain data are 4D volumes
(x,y,z,time); the vectors associated with the principal strains are
5D volumes (x,y,z,time,[e_x e_y e_z]).
The images are all oriented in RAI => +x is right to left, +y is
anterior to posterior, +z is inferior to superior. Included is a
toolbox to load NIFTI files (.nii) into Matlab (niftimatlib-1.2).
An example of how to use the toolbox is given below:
First, unzip the *gz files using gunzip in Matlab or similar:
gunzip('mN3978_1_rot_T1.nii.gz');
N = nifti('mN3978_1_rot_T1.nii');
T1 = N.dat(:,:,:); % 3D matrix with signal intensity of the image
at each voxel
mat = N.mat; % qform matrix
res = N.hdr.pixdim(2:4); % image resolution
clear N % always clear N once you've pulled the necessary
information. Otherwise, possible to overwrite data
N = nifti('N3978_1_rot_u.nii');
u = N.dat(:,:,:,:); % number of colons should equal the dimension
of the data. If unsure, N.dat.dim provides the data dimension
clear N
When the data is loaded in, dimension 1 is x, dimension 2 is y, and
dimension 3 is z. I often permute the first and second dimensions
when visualizing the images so that the x-dimension is along the
x-axis in Matlab.
Changes:
Includes DTI Data.
|