surface-tutorial

Description

Cortical Surface Morphometry Tutorial1 - Prepare data Convert images from DICOM to Nifti :• open the dicom image with Anatomist (one file per slice in DICOM, you just need to open the first one (name ends with .1). In the file selction box you need to chose the "dicom" filter otherwise you will not see it (default in Anatomist is another format). • Right click on the dataset in Anatomist main window and chose "file"->"save".• Save the file with a '.nii' extension. It is the Nifti format, this is the best one to use. If orientation is 'bottom to top' (upside down in sagittal view, see image), you need to flip it :AimsFlip -i image.nii -o image_flipped.nii -m ZZ'bottom to top' image and the 'top to bottom' flipped versionNote : before running Aims commands, such as AimsFlip, you might need to run another command : brainvisa_setup. Open a command window, type 'brainvisa_setup' and then all Aims commands will be available in this command window.2 - Import the Nifti file in the Brainvisa databaseThe flipped image needs to be imported in a Brainvisa database. Start Brainvisa and run the 'Data Management -> import -> T1 MRI -> Import T1 MRI' process. Choose the proper protocol and subject name. 3 - Run the T1 segmentation pipelineThis pipeline segment all tissues, compute all meshes and extract the sulci graph. Open it in 'T1 MRI'->'Segmentation Pipeline'->'T1 Pipeline 2007'Select the subject T1 image with the green database browser button and all other ...

Informations

• Publié par: Phoim
• Nombre de lectures: 18
• Langue: English

Extrait

"save".• Save the file with a '.nii' extension. It is the Nifti format, this is the best one to use. If orientation is 'bottom to top' (upside down in sagittal view, see image), you need to flip it :AimsFlip -i image.nii -o image_flipped.nii -m ZZ'bottom to top' image and the 'top to bottom' flipped versionNote : before running Aims commands, such as AimsFlip, you might need to run another command : brainvisa_setup. Open a command window, type 'brainvisa_setup' and then all Aims commands will be available in this command window.2 - Import the Nifti file in the Brainvisa databaseThe flipped image needs to be imported in a Brainvisa database. Start Brainvisa and run the 'Data Management -> import -> T1 MRI -> Import T1 MRI' process. Choose the proper protocol and subject name. 3 - Run the T1 segmentation pipelineThis pipeline segment all tissues, compute all meshes and extract the sulci graph. Open it in 'T1 MRI'->'Segmentation Pipeline'->'T1 Pipeline 2007'Select the subject T1 image with the green database browser button and all other ..." />

1 - Prepare data

Cortical Surface Morphometry Tutorial

Convert images from DICOM to Nifti :

• open the dicom image with Anatomist (one file per slice in DICOM, you just need to open the first one (name ends with .1). In the file selction box you need to chose the "dicom" filter otherwise you will not see it (default in Anatomist is another format). • Right click on the dataset in Anatomist main window and chose "file"->"save". • with a '.nii' extension. It is the Nifti format, this is the best one to use.Save the file

If orientation is 'bottom to top' (upside down in sagittal view, see image), you need to flip it :

AimsFlip -i image.nii -o image_flipped.nii -m ZZ

'bottom to top' image and the 'top to bottom' flipped version

Note : before running Aims commands, such as AimsFlip, you might need to run another command : brainvisa_setup. Open a command window, type 'brainvisa_setup' and then all Aims commands will be available in this command window.

2 - Import the Nifti file in the Brainvisa database

The flipped image needs to be imported in a Brainvisa database. Start Brainvisa and run the'Data Management -> import -> T1 MRI -> Import T1 MRI' process. Choose the proper protocol and subject name.

3 - Run the T1 segmentation pipeline

This pipeline segment all tissues, compute all meshes and extract the sulci graph. Open it in'T1 MRI'->'Segmentation Pipeline'->'T1 Pipeline 2007' Select the subject T1 image with the green database browser button and all other fields will be automatically filled except those for the the AC, PC, inter-hemispheric point and left hemisphere point.

The T1 segmentation pipeline

Click on the anatomist icon (see figure) next to the AC field. Anatomist is launched with the T1 image opened. Find AC, click on it, then click again on the anatomist button. The field is filled with the associated coordinate values. Then find PC, click on it, then click on the anatomist icon next to the PC field. The field is then filled in.Then go towards the top of the brain (higher than AC and PC) and click somewhere in the inter- hemispheric plane. Then click on the anatomist icon next to the ´Inter-hemispheric point' filled. Finally click somewhere in the middle of the left hemisphere, then click on the Anatomist icon next to the 'Left hemisphere point' field.Caution :unless you know otherwise, your image will be displayed in Radiological convention in Anatomist, so the left hemisphere is on the right side.

Four point to select (top to bottom, left to right) : AC, PC, inter-hemispheric point, left hemisphere point

Troubleshooting : if the pipeline fails (it happens), it is often because of the histogram analysis process (even if it fails after this process). The first thing to try is to select the histogram analysis component and change the 'undersampling' field value from 'Auto' to '16'.

The pipeline produces a number of images, meshes, and sulci graphs. It is very important that the white matter mesh is of good enough quality (see figure) since it is the one on which all further surface-based processings are performed. The white matter mesh is named sujet_Lwhite.mesh or sujet_Rwhite.mesh

white matter mesh

4 – Run the automatic sulci labeling tool

The process is not the one shown at the end of the T1 segmentation pipeline (ticked off by default). You should open the one in ´Morphometry'->'Parallel recognition'. This process run the recognition 10 times and keeps the best result, thus increasing robustness.

Sulci recognition

The process takes the unlabeled graph (Lsujet.arg or Rsujet.arg) and generates a labeled graph

(Lsujet_default_session_auto.arg).

5 – Run the cortical parameterization tool

The process is in the Cortical Surface toolbox. Everything you need to know about it is available on a web page :

http://olivier.coulon.perso.esil.univmed.fr/Cortical_surface_parameterization.html

6 – Tools and other little things

According to the Brainvisa database organization, there is aprotocol/directory in which there is one directory per subject.

Inprotocol/subject/there is at1mri/directory that contains the result of all processings of the t1 image (images, meshes, graph) and asurface/directory that contains the result of the parameterization.

Coordinate grid : you can generate a coordinate grid on the surface. No Brainvisa process for that yet (soon) but a simple command line. Go in thesubject/surface/directory and type (possibly after a 'brainvisa_setup' command) :

AimsCoordinateGridMesh -m ../t1mri/default_acquisition/default_analysis/segmentation/mesh/sujet_Lwhite.mesh -x sujet_L_lat.tex -y sujet_L_lon.tex -c r -o sujet_L_grid_r.mesh

The '-c' argument corresponds to the type of grid you want : '-c r' for a regular grid (see figure), and '-c c' for the grid that defines the parcellation. The '-o' argument defines the output grid file name. The grid is a mesh that you can put in a 3d window with the surface using Anatomist. The name is completely free.

Coordinate grids : the gyri grid, in grey, superiposed with the gyri, and a regular grid, in black

Mesh Inflation : it is a good idea to look at the grid on an inflated version of the mesh, for visualization purpose. To do this you have to inflate the mesh with the surface inflation tool in : 'T1MRI'->'Surface'. The default parameters perform a 'full' inflation. It is recommended to change the 'iterations' parameter from 500 (the default value) to 80, such that the main folds are preserved, which makes it easier for visualization.

You can then generate the grids for the inflated mesh with a command like :

AimsCoordinateGridMesh -m ../t1mri/default_acquisition/default_analysis/segmentation/mesh/sujet_Lwhite_inflated.mesh -x sujet_L_lat.tex -y sujet_L_lon.tex -c r -o sujet_L_grid_inflated_r.mesh

Grids on an inflated cortex