STEAM Logo
STEAM
Statistical Template Estimation for Abnormality Mapping


Download Example STEAM Templates

This STEAM preterm infant DTI template collection is provided under the Creative Commons Attribution-NonCommercial-ShareAlike license. It is provided "as is" without any implied warranties for any particular purpose. In no event shall the authors of this site be liable for any damage or tort arising from the use of these template, even if we are advised of the possibility of such damage.

If you used these STEAM templates in a publication, we ask that you cite the appropriate source (i.e. this website) as follows:

B. G. Booth, S. P. Miller, C. J. Brown, K. J. Poskitt, V. Chau, R. E. Grunau, A. R. Synnes, and G. Hamarneh. "STEAM - Statistical Template Estimation for Abnormality Mapping: a Personalized DTI Analysis Technique with Applications to the Screening of Preterm Infants", NeuroImage, Volume 125, pp. 705-723, 2016.

To download the STEAM templates, please select the statistical templates from the full collection below. The selected templates will be collected in a zip file and an email with a link to the zip file will sent soon after. Each image in the STEAM template collection is provided in compressed nifti format. Full tensor images are provided in upper triangular order ([dxx dxy dxz dyy dyz dzz]).


Diffusion Feature

Post-Menstrual Age Groups

28-31 Weeks

32-36 Weeks

37-40 Weeks

41-45 Weeks











Full Tensor

what is this?

Full Tensor

The full diffusion tensor is a 3x3 matrix with six unique elements. The diffusion tensor can also be thought of as the co-variance matrix of a normal distribution that captures the probability of water molecule displacements. The tensor images provided here are given in log-Euclidean space to ensure that any tensor arithmetic is performed correctly. For more information on log-Euclidean space and tensors, please see the original paper on the topic.


Fractional Anisotropy

what is this?

Fractional Anisotropy

Fractional Anisotropy (FA) is a measure of the variance in diffusion rates as the direction of the diffusion changes. High FA values are typically associated with fibrous tissue like the white matter of the brain, while the FA at a voxel is zero when diffusion at that voxel is equal in all directions. For more information on how FA is computed, please see this paper.


Mean Diffusivity

what is this?

Mean Diffusivity

Mean Diffusivity (MD) is a measure of how much diffusion is occurring at a given voxel. Larger MD implies less cell structure while smaller MD implies the presence of significant cell structure. For more information on how MD is computed, please see this paper.


Primary Eigenvalue

what is this?

Primary Eigenvalue

The primary eigenvalue (L1) is the largest eigenvalue of the diffusion tensor and is also referred to as axial diffusivity. In fibrous tissue, L1 indicates the rate of diffusion along the fibers at a voxel.


Secondary Eigenvalue

what is this?

Secondary Eigenvalue

The secondary eigenvalue (L2) is the second-largest eigenvalue of the diffusion tensor. In fibrous tissue, L2 indicates the rate of diffusion perpendicular to the fibers at a voxel.


Tertiary Eigenvalue

what is this?

Tertiary Eigenvalue

The Tertiary eigenvalue (L3) is the second-largest eigenvalue of the diffusion tensor. In fibrous tissue, L3 indicates the rate of diffusion perpendicular to the fibers at a voxel.


Radial Diffusivity

what is this?

Radial Diffusivity

Radial Diffusivity (RD) is the measure of the diffusion rate perpendicular to the axonal fiber direction. RD is computed as the average of the two smallest eigenvalues of the diffusion tensor. An abnormally high value of RD may indicate axonal dematuration.


Linear Diffusion Coefficient

what is this?

Linear Diffusion Coefficient

The linear diffusion coefficient (CL) captures what fraction of the diffusion at a voxel is confined to a single direction (or line). Large CL values would be found in fibrous tissue where all the axons are aligned in the same direction. For more information on how the linear diffusion coefficient is computed, please see this paper.


Planar Diffusion Coefficient

what is this?

Planar Diffusion Coefficient

The planar diffusion coefficient (CP) captures what fraction of the diffusion at a voxel is confined to a single plane. Large CP values would be found in fibrous tissue where there are two axon populations that are crossing. For more information on how the planar diffusion coefficient is computed, please see this paper.


Spherical Diffusion Coefficient

what is this?

Spherical Diffusion Coefficient

The spherical diffusion coefficient (CS) captures how isotropic, or spherical, the diffusion profile is. Larges CS values indicate that the axons at a particular voxel are not coherently aligned in a single direction. For more information on how the spherical diffusion coefficient is computed, please see this paper.


Relative Anisotropy

what is this?

Relative Anisotropy

Relative anisotropy (RA) is a measure of the variance in diffusion rates as the direction of the diffusion changes. It is computed in a similar fashion to fractional anisotropy and is known to have a lower signal-to-noise ratio. Like fractional anisotropy, large RA values occur in fibrous tissue while smaller RA values occur in areas where there is no preferred direction of diffusion. For more information on how relative anisotropy is computed, please see this paper.


log-Fractional Anisotropy

what is this?

log-Fractional Anisotropy

log-Fractional Anisotropy (LFA) is the fractional anisotropy computed from the log-Euclidean diffusion tensor. LFA behaves in a similar way to fractional anisotropy: large LFA values occur in fibrous tissue while smaller LFA values occur in areas where there is no preferred direction of diffusion. For more information on log-Euclidean tensors, please see the original paper on the topic. For more information on how fractional anisotropy is computed, please see this paper.


Tensor Norm

what is this?

Tensor Norm

The tensor norm (VOL) is exactly what it sounds like: the Frobenious norm of the diffusion tensor. Like mean diffusivity, VOL tells us how much diffusion is occurring at a given voxel. Larger VOL implies less cell structure while smaller VOL implies the presence of significant cell structure.


Volume Ratio

what is this?

Volume Ratio

Volume ratio (VR) is a measure of how spherical, or isotropic, the diffusion profile is at a given voxel. A high measure of VR suggests that the volume of the diffusion tensor ellipsoid is very close to the volume of a sphere with a diameter equal to the average length of the ellipsoid axes. More information on the volume ratio can be found in this paper





Mean Images (co-) Variance Images Normalcy p-value Images




Spatial Scales

what is this?

Spatial Scales

We provide each template with different levels of smoothing applied to the images from which the templates were generated. We do this to allow our templates to fit into a voxel-based analysis pipeline. It is general practice to smooth images before performing any voxel-based analysis and the reasons behind doing so are given in our paper.

0 mm

1 mm

2 mm

3 mm

4 mm

5 mm

6 mm

7 mm

8 mm



(scales listed by Full Width at Half Maximum values)



Name: Research interests and other comments:
Position:
Affiliation:
Email: * (link to download will be sent to this email)

 


© SFU Medical Image Analysis Lab, 2014