Differences between revisions 25 and 59 (spanning 34 versions)

FEM tensors estimation with BrainSuite

Authors: Takfarinas Medani, Francois Tadel, Anand Joshi and Richard Leahy

[TUTORIAL UNDER WRITING: NOT READY FOR PUBLIC USE]

In this tutorial, we describe the process of the estimation of the realistic conductivity tensors for the brain tissues using the BrainSuite software. The main purpose is to generate the conductivity tensors for the FEM computation as introduced on this page.

The realistic tensors are estimated from the Diffusion-Weighted Images (DWI). For this purpose, Brainstorm calls internally the BrainSuite software to compute the diffusion tensors on each brain MRI voxel. Afterward, the Effective Medium Approach (EMA) is applied to convert the diffusion tensors to the conductivity tensors.

The following tutorial explains to the users how to generate conductivity tensors with BrainSuite from the Brainstorm graphical interface.

Further documentation about the previous usage of BrainSuite within Brainstorm can be found in this page.

Contents

Requirement
Brainsuite Installation
Dataset
Realistic condctivity tensors
Artificial/simulated conductivity tensors
References

Requirement

You have already followed all the introduction tutorials
You have a working copy of Brainstorm installed on your computer
For the DWI data, only the NIfTi files are supported

Brainsuite Installation

Download the latest version of BrainSuite from http://www.brainsuite.org/download.
Install it on your computer by following the instructions in BrainSuite's quick start installation guide.
Note that you will be using BrainSuite Diffusion Pipeline(BDP), so you need to install a compatible MATLAB Compiler Runtime(last version).
Start BrainSuite to check if the installation (It's not required to open BrainSuite to run this tutorial).
The BrainSuite installation folder should be informed in the Brainstorm preferences

Dataset

In this tutorial, we use the Brainsuite dataset example available on the BrainSuite tutorial webpage. User can also download directly these data from these links: MRI T1w and MRI DWI

The first file contains the T1 MRI data, with the name '2523412.nii.gz', as in this figure,

The second folder contains the diffusion data and should have at least three files,

Where the *.bval is a text file that contains the value of the gradient, and the *.bvec is also a text file that contains the orientation of the gradient. The nii.gz file is the NifTi file where the DWI where the images are stored.

Realistic condctivity tensors

Load the T1 MRI data to brainstorm

First, you need to create a new subject in your protocol, let call it the 'BrainSuiteSubject'. Then import the T1 MRI of the subject and set the fiducials points as explained in the previous tutorial.

The T1 MRI is required since the BDP uses the T1 space for its computation. Furthermore, this is required since it will be used to align the tensors to the FEM mesh later.

Diffusion tensor generation (DTI) from DWI

In this step, Brainstorm calls the Brainsuite internally, and the diffusion tensors are computed.

Right-click on the subject and then select the item "Convert DWI to DTI".

Then follow the popup windows by selecting the DWI, you may need to extract the zip file before.

If the bval, and bval files are in the same folder, Brainstorm will detect them automatically, otherwise, the user will be asked to browse the files one by one (as it's the case in this tutorial).

Brainstorm calls internally the BrainSuite process, and compute the diffusion tensors.

At the end of this process, a new node will appear in the Brainstorm database with the name 'DTI-EIT'. This name refers to, DTI: diffusion tensors images, and EIG for eigenvalue, since the eigenvalues and eigenvectors are computed at voxel and stored in Brainstorm database.

If you check the structure of the file DTI-EIG, by right click -> File and then 'Display file contents', the following figure is displayed.

The size of the matrix is 128x256x256x12, where the first 3 values are the same as the size of the T1 MRI and 12 corresponds to the 3 eigenvectors components (9) and eigenvalues (3)

Conductivity tensor generation from DTI

The Effective Medium Approach is applied to convert the diffusion tensors to the conductivity tensors.

FEM mesh head model

This step requires the FEM mesh of the head model. You can generate the FEM head model from the MRI data as explained on this page.

For the following, we used the SimNibs FEM mesh generation. The following figure shows the FEM mesh obtained with the SimNibs method using the T1 MRI.

Note that this mesh is obtained only from the T1, the use of the T2 is highly recommended if it's available, as recommended in the FEM mesh tutorial.

Computation of FEM mesh tensors

Once the FEM mesh and the DTI tensors are available in the Brainstorm database, the next step for the FEM tensors can be performed by the following:

- Right-click on the FEM mesh - Compute FEM tensors

Brainstorm checks the available tissues in the FEM head model and displays the following panel

This panel lists the tissues available in the FEM head model and assigns a default value of the conductivity for each compartment. Users can change these values to their own if needed.

DTI values can be used to generate conductivity tensors for the white matter (and in some cases for the grey matter). Please, note that the DWI can be used only for the brain tissues and not for the outers compartments (skull and skin)

In this tutorial (and in most cases) we select the white matter. Select the WM anisotropy and keep all the other tissues as isotropic, then these additional options appear asking for the method to use.

The available methods are:

- Effective Medium approach (EMA)

- Effective Medium approach with volume constraints (EMA + VC)

- Simulated or the artificial anisotropy

Only the two first methods require the DTI. More information about these methods can be found on these references [ref1][ref2] and in our main paper [link]

In this tutorial, we use the method "EMA + VC", where the final tensors are constrained to fits the volume of the equivalent isotropic tensor volume.

Visulation of FEM mesh tensors

Once the FEM tensors are successfully computed, they are stored in the FEM head node. By right-clicking on the FEM head, new menu items are added that gives the possibilities to display the FEM tensors either as ellipsoids or as vectors in the direction of the main eigenvector.

The tensors can be displayed either on the FEM mesh or overlaid on the MRI. The following figures show an example of the obtained tensors displayed on the white matter.

On the left, the tensors as a line on the direction of the main eigenvector. On the right, the tensors displayed as ellipsoids. The orientation of the tensor is color-coded as follows: red for right-left, green for anterior-posterior, and blue for superior-inferior.

Note that the quality of the tensors depends on the DWI data and the number of acquisition direction.

Users can also display the tensors on specific tissues, for example on the white matter (left figure) or overlay on the MRI (right figure).

Recommendation

In the case where the user wants to use generate isotropic tensors, then the DTI is not required. For that case, keep all the options to 'isotropic', the recommended display is the 'Ellipsoids', and the final shape will be a sphere (isotropic direction). In the case where more than one FEM head model is in the database, the highlighted one in the green color will be used.

Advanced

Artificial/simulated conductivity tensors

In the case where the DWI is not available, or in the case where the users desire to evaluate the effect of the conductivity change on the head model, the artificial conductivity can be used.

Users can reach this option by following this tutorial and select the third method in this panel.

Two approaches are integrated within Brainstorm. Either Wang's constraint or the volume's constraint (Wolters). The common feature between these methods is the ratio between the transversal and longitudinal conductivity ratio.

A common example is the skull anisotropy simulation, where the longitudinal conductivity can be higher than the transversal conductivity, the ratio can vary from 2 to 10 [ref]. In this tutorial, we keep all the tissue as isotropic, except the skull, we use a ratio of 0.1 and select the volume constraint. The following figures show the results of this example.

"eigenvalues parallel (longitudinal) and perpendicular (transverse) to the fiber directions" for 1:10 anisotropy (transverse:longitudinal)

References

To be completed soon

===TODO===

Check the error in the simnibs mesh in X direction and overlay on mri check the error with the brain2mesh Correct the ratio from integer to float check the meaning of transversal/longitidunal in the code add an interactive way yo change the size of the tensor.. important correct the name of the simulated method, correct the EMC and remove the VC and change the coefficcient

-  ⇤ ← Revision 25 as of 2020-08-25 02:48:28 → 
  Size: 8682
  Editor: TakfarinasMedani
  Comment:
+   ← Revision 59 as of 2020-08-27 07:03:21 → ⇥
  Size: 11339
  Editor: TakfarinasMedani
  Comment:
-Deletions are marked like this.
+Additions are marked like this.
 Line 8:
-Describe brainsuite here : here we will describe the process of the brain tissues anisotrpy estimation and the different functions that brainstorm offers.
+In this tutorial, we describe the process of the estimation of the realistic conductivity tensors for the brain tissues using the [[http://brainsuite.org/|BrainSuite software]].   The main purpose is to generate the conductivity tensors for the FEM computation as introduced on [[https://neuroimage.usc.edu/brainstorm/Tutorials/Duneuro|this page]].
 Line 10:
-This tutorial explains how to use Brainsuite to estimate the anisotropy of the brain tissues.
+The realistic tensors are estimated from the Diffusion-Weighted Images (DWI). For this purpose, Brainstorm calls internally the BrainSuite software to compute the diffusion tensors on each brain MRI voxel. Afterward, the Effective Medium Approach (EMA) is applied to convert the diffusion tensors to the conductivity tensors.
 Line 12:
-refer to this page
+The following tutorial explains to the users how to generate conductivity tensors with BrainSuite from the Brainstorm graphical interface.
 Line 14:
-[[https://neuroimage.usc.edu/brainstorm/Tutorials/SegBrainSuite?highlight=(Anand)|https://neuroimage.usc.edu/brainstorm/Tutorials/SegBrainSuite?highlight=%28anand%29]]

The realistic tensors are estimated from the Diffusion-Weighted Images (DWI). For this purpose, Brainstorm calls internally the BrainSuite Diffusion Pipeline to compute the diffusion tensors on each brain voxel. Afterward, the Effective Medium Approach is applied to convert the diffusion tensors to the conductivity tensors. The following section shows the users how to do it from the graphical interface.

Only the NIfTI is supported. All the diffusion data, including the DWI file and direction and the value of the gradient files, respectively the *.nii, the *.bval and the *.bvec are required. Ideally, these files should have the same name and saved in the same folder.
+Further documentation about the previous usage of BrainSuite within Brainstorm can be found in this [[https://neuroimage.usc.edu/brainstorm/Tutorials/SegBrainSuite|page]].
-Line 25:
+Line 21:
+ * For the DWI data, only the NIfTi files are supported
-Line 28:
+Line 25:
-Line 29:
+Line 27:
-Line 30:
+Line 29:
 Line 31:
-Line 37:
+Line 38:
-In this tutorial, we use the Brainsuite dataset example available on the Brainsuite tutorial webpage
+In this tutorial, we use the Brainsuite dataset example available on the BrainSuite  [[http://brainsuite.org/tutorials/|tutorial webpage]]. User can also download directly these data from these links: [[http://brainsuite.org/WebTutorialData/BrainSuiteTutorialSVReg_Sept16.zip|MRI T1w]] and [[http://brainsuite.org/WebTutorialData/DWI_Feb15.zip|MRI DWI]]
-Line 39:
+Line 40:
-http://brainsuite.org/tutorials/. The T1w of the subject can be download from this link  [[http://brainsuite.org/WebTutorialData/BrainSuiteTutorialSVReg_Sept16.zip|BrainSuiteTutorialSVReg_Sept16.zip]]. The DWI from this link  [[http://brainsuite.org/WebTutorialData/DWI_Feb15.zip|DWI_Feb15.zip]]
+The first file contains the T1 MRI data, with the name '2523412.nii.gz', as in this figure,
-Line 41:
+Line 42:
-The first link contains the T1 MRI, with the name '2523412.nii.gz'. as in this figure,
+{{attachment:vierMriFolder.JPG||height="150",width="650"}}
-Line 43:
+Line 44:
-{{attachment:vierMriFolder.JPG||height="130",width="580"}}
+The second folder contains the diffusion data and should have at least three files,
-Line 45:
+Line 46:
-The second file is the DWI and should contain at least three files
+{{attachment:vierDWIFolder.JPG||height="180",width="650"}}
-Line 47:
+Line 48:
-{{attachment:vierDWIFolder.JPG||height="150",width="580"}}

Where the *.bval is a text file that contains the value of the gradient, and the *.bvec is also a text file that contains the orientation of the gradient. The nii.gz file is the NifTi file of the DWI where the images are stored.
+Where the *.bval is a text file that contains the value of the gradient, and the *.bvec is also a text file that contains the orientation of the gradient. The nii.gz file is the NifTi file where the DWI where the images are stored.
-Line 53:
+Line 52:
-First, you need to create a new subject in your protocol, let call it the 'BrainSuiteSubject'. Then import the T1 MRI of the subject and set the fiducials points as explained in the previous tutorial.
+First, you need to create a new subject in your protocol, let call it the 'BrainSuiteSubject'. Then import the T1 MRI of the subject and set the fiducials points as explained in the [[https://neuroimage.usc.edu/brainstorm/Tutorials/ImportAnatomy|previous tutorial]].
-Line 55:
+Line 54:
-The T1 MRI is required since all the BDP uses the T1 space for its computation. Furthermore, this is required since it will be used to align the tensors to the FEM mesh later.
+The T1 MRI is required since the BDP uses the T1 space for its computation. Furthermore, this is required since it will be used to align the tensors to the FEM mesh later.
-Line 62:
+Line 61:
-Right-click on the subject and then select the item "Convert DWI to DTI". Then follow the popup windows by selecting the DWI, bval, and bval files. If these files are in the same folder, Brainstorm will detect them automatically, otherwise, the user will be asked to browse the files one by one (as it's the case in this tutorial).
+Right-click on the subject and then select the item "Convert DWI to DTI".

{{attachment:dwi2dti.jpg||height="300",width="550"}}

Then follow the popup windows by selecting the DWI, you may need to extract the zip file before.

{{attachment:importDWI.JPG||height="280",width="650"}}

If the bval, and bval files are in the same folder, Brainstorm will detect them automatically, otherwise, the user will be asked to browse the files one by one (as it's the case in this tutorial).

Brainstorm calls internally the BrainSuite process, and compute the diffusion tensors.
-Line 66:
+Line 75:
-{{attachment:mriAndDTI-EIG.JPG||height="100",width="250"}}
+{{attachment:mriAndDTI-EIG.JPG||height="100",width="230"}}
-Line 70:
+Line 79:
-{{attachment:EIG-hardDisc.JPG||height="100",width="250"}}
+{{attachment:EIG-hardDisc.JPG||height="200",width="450"}}
-Line 75:
+Line 84:
 The Effective Medium Approach is applied to convert the diffusion tensors to the conductivity tensors.
-Line 78:
+Line 87:
-This step requires the FEM mesh of the head model. You can generate the FEM head model from the MRI as explained on this page (add link).
+This step requires the FEM mesh of the head model. You can generate the FEM head model from the MRI data as explained on [[https://neuroimage.usc.edu/brainstorm/Tutorials/FemMesh|this page]].
-Line 80:
+Line 89:
-For the following, we used the SimNibs FEM mesh generation. The following figure shows the FEM mesh obtained from using the T1 MRI.
+For the following, we used the SimNibs FEM mesh generation. The following figure shows the FEM mesh obtained with the SimNibs method using the T1 MRI.
-Line 82:
+Line 91:
-{{attachment:Mri&femMeshView.JPG|Mri&femMeshView.JPG|height="250",width="230"}} {{attachment:femMeshView.JPG||height="250",width="250"}}
+{{attachment:Mri&femMeshView.JPG|Mri&femMeshView.JPG|height="300",width="260"}} {{attachment:femMeshView.JPG||height="300",width="280"}}
-Line 84:
+Line 93:
-Note that this mesh is obtained only from the T1, the use of the T2 is highly recommended if it's available.
+Note that this mesh is obtained only from the T1, the use of the T2 is highly recommended if it's available, as recommended in the [[https://neuroimage.usc.edu/brainstorm/Tutorials/FemMesh|FEM mesh tutorial]].
-Line 86:
+Line 95:
-==== FEM mesh tensors ====
+==== Computation of FEM mesh tensors ====
-Line 88:
+Line 97:
-  - Right-click on the FEM mesh
- Compute FEM tensors
-Line 92:
+Line 98:
-{{attachment:menuGenerateFemTensors.png||height="250",width="250"}}
+- Right-click on the FEM mesh - Compute FEM tensors

{{attachment:menuGenerateFemTensors.png||height="280",width="250"}}
-Line 96:
+Line 104:
-{{attachment:FEMConductivitiesIsoPanel.JPG||height="250",width="250"}}
+{{attachment:FEMConductivitiesIsoPanel.JPG||height="220",width="250"}}
-Line 98:
+Line 106:
-This panel lists the tissues available in the FEM head model and assign a default value of the conductivity for each compartment. Users can change these values to their own if needed.
+This panel lists the tissues available in the FEM head model and assigns a default value of the conductivity for each compartment. Users can change these values to their own if needed.
-Line 100:
+Line 108:
-DTI values can be used to generate conductivity tensors for the white matter (and in some cases for the grey matter). To do so, select the anisotropy button for the white matter and then these additional options appear
+DTI values can be used to generate conductivity tensors for the white matter (and in some cases for the grey matter). Please, note that the DWI can be used only for the brain tissues and not for the outers compartments (skull and skin)
-Line 102:
+Line 110:
-{{attachment:FEMConductivitiesAnisoPanel.JPG||height="250",width="250"}}
+In this tutorial (and in most cases) we select the white matter. Select the WM anisotropy and keep all the other tissues as isotropic, then these additional options appear asking for the method to use.
-Line 105:
+Line 112:
+{{attachment:FEMConductivitiesAnisoPanel.JPG||height="300",width="250"}}
-Line 106:
+Line 114:
-In the case where the user wants to use generate isotropic tensors, then the DTI is not required.
+The available methods are:
-Line 108:
+Line 116:
+- Effective Medium approach (EMA)
-Line 109:
+Line 118:
+- Effective Medium approach with volume constraints  (EMA + VC)
-Line 110:
+Line 120:
+- Simulated or the artificial anisotropy
-Line 111:
+Line 122:
+Only the two first methods require the DTI. More information about these methods can be found on these references [ref1][ref2] and in our main paper [link]
-Line 112:
+Line 124:
-In the case where more than one FEM head model is in the database, the highlighted one in the green color will be used.
+In this tutorial, we use the method "EMA + VC", where the final tensors are constrained to fits the volume of the equivalent isotropic tensor volume.
-Line 114:
+Line 126:
-The FEM head model to use for tensors should be selected and highlighted with the green color (double click on the FEM mesh node to select it)
+==== Visulation of FEM mesh tensors ====
Once the FEM tensors are successfully computed, they are stored in the FEM head node. By right-clicking on the FEM head, new menu items are added that gives the possibilities to display the FEM tensors either as ellipsoids or as vectors in the direction of the main eigenvector.
-Line 116:
+Line 129:
-When this is done, then right-click on the subject -> 'Convert DWI to DTI',
+{{attachment:menuDisplayTensors.jpg||height="300",width="250"}}
-Line 118:
+Line 131:
-Brainstorm will load the available tissue in the FEM head model and the following windows appear.
+The tensors can be displayed either on the FEM mesh or overlaid on the MRI. The following figures show an example of the obtained tensors displayed on the white matter.
-Line 120:
+Line 133:
-Select the WM anisotropy and kee all the other tissues as isotropic.
+{{attachment:meshViewTensorsLines.JPG||height="300",width="350"}} {{attachment:meshViewTensorsTensorsTops.JPG||height="300",width="350"}}
-Line 122:
+Line 135:
-The process of conversion from DWI to Conductivity tensors uses the EMA, furthermore, brainstorm proposes the option to use the adaptative EMA with the volume constraint option [ref]. In this example, we select the EMA with the VC.
+On the left, the tensors as a line on the direction of the main eigenvector. On the right, the tensors displayed as ellipsoids. The orientation of the tensor is color-coded as follows: red for right-left, green for anterior-posterior, and blue for superior-inferior.
-Line 124:
+Line 137:
-The process will take around 10 min, and then the FEM tensors are computed and stored in the FEM structure. [explain how it is organized and how to use it outside brainstorm ]
+Note that the quality of the tensors depends on the DWI data and the number of acquisition direction.
-Line 126:
+Line 139:
-Brainstorm will load the available tissue in the FEM head model and the following windows appear.
+Users can also display the tensors on specific tissues, for example on the white matter (left figure) or overlay on the MRI (right figure).
-Line 128:
+Line 141:
-Select the WM anisotropy and kee all the other tissues as isotropic.
+{{attachment:meshViewTensorsLinesWM2.JPG||height="300",width="350"}} {{attachment:tensorsOnMri.JPG||height="300",width="300"}}
-Line 130:
+Line 143:
-The process of conversion from DWI to Conductivity tensors uses the EMA, furthermore, brainstorm proposes the option to use the adaptative EMA with the volume constraint option [ref]. In this example, we select the EMA with the VC.
+==== Recommendation ====
In the case where the user wants to use generate isotropic tensors, then the DTI is not required. For that case, keep all the options to 'isotropic', the recommended display is the 'Ellipsoids', and the final shape will be a sphere (isotropic direction).   In the case where more than one FEM head model is in the database, the highlighted one in the green color will be used.
-Line 132:
+Line 146:
-The process will take around 10 min, and then the FEM tensors are computed and stored in the FEM structure. [explain how it is organized and how to use it outside brainstorm ]
+<<TAG(Advanced)>>
-Line 135:
+Line 149:
-In the case where the DWI is not available, or the users desire to evaluate the effect of the conductivity change on the model, the artificial conductivity can be used.
+In the case where the DWI is not available, or in the case where the users desire to evaluate the effect of the conductivity change on the head model, the artificial conductivity can be used.
-Line 137:
+Line 151:
-Two approaches are integrated within Brainstorm. Either the
+Users can reach this option by following this tutorial and select the third method in this panel. {{attachment:artificialTensors.JPG||height="450",width="280"}}

Two approaches are integrated within Brainstorm. Either Wang's constraint or the volume's constraint (Wolters). The common feature between these methods is the ratio between the transversal and longitudinal conductivity ratio.

A common example is the skull anisotropy simulation, where the longitudinal conductivity can be higher than the transversal conductivity, the ratio can vary from 2 to 10 [ref]. In this tutorial, we keep all the tissue as isotropic, except the skull, we use a ratio of 0.1 and select the volume constraint. The following figures show the results of this example.

"eigenvalues parallel (longitudinal) and perpendicular (transverse) to the fiber directions" for 1:10 anisotropy (transverse:longitudinal)

{{attachment:skullAniso.JPG||height="250",width="300"}} {{attachment:skullAniso2.JPG||height="250",width="300"}}
-Line 140:
+Line 162:
+To be completed soon

===TODO===

Check the error in the simnibs mesh in X direction and overlay on mri check the error with the brain2mesh  Correct the ratio from integer to float  check the meaning of transversal/longitidunal in the code add an interactive way yo change the size of the tensor.. important   correct the name of the simulated method,   correct the EMC and remove the VC and change the coefficcient

Software

Users

Development