Beauchamp:Creating a Surface Model and Electrode Localization (by Muge Ozker Sertel)

Also see: Beauchamp:Electrode Localization and Naming

Step 1: Get the MR and CT scans of the patient

Step 2: Create a patient folder on the server i.e. /Volumes/data/UT/YAR (YAR is the patient ID in this example case) Copy in MR and CT scans with OsiriX, then export the images to the patient's folder. Choose Hierarchical and Decompress DICOMs options during the export. Take out the spaces in MR and CT folder names. The folder names should look like as follows:

/Volumes/data/UT/YAR/LastName_FirstName_MR/MRBRAINWITHOUTCONTRAST
/Volumes/data/UT/YAR//LastName_FirstName_CT/CTBRAINWITHOUTCONTRAST

Step 3: Process anatomical MR data:

In terminal window type:

set ec = YAR 
set session = /Volumes/data/UT/YAR/afni 

This way the processed anatomical files will be saved in the afni folder under the patient's directory.

There should be two different T1 scan folders in the MRBRAINWITHOUTCONTRAST folder. These folders are usually named T1_TFE_NSA01 (include images from 1st T1 scan) and T1_TFE_NSA02 (include images from 2nd T1 scan). We use the images in these folders.

• Create 3D datasets for use with AFNI from these 2D MR image files

cd /Volumes/data/UT/YAR/LastName_FirstName_MR/MRBRAINWITHOUTCONTRAST/T1_TFE_NSA01
to3d -session $session -prefix {$ec}anatr1 IM-0011-0*  

cd /Volumes/data/UT/YAR/LastName_FirstName_MR/MRBRAINWITHOUTCONTRAST/T1_TFE_NSA02
to3d -session $session -prefix {$ec}anatr2 IM-0012-0* 

• Align the 1st T1 to the 2nd T1

This takes approximately 7 minutes.

cd /Volumes/data/UT/YAR/afni (YARanatr1 and YARanatr2 are in afni folder)
3dAllineate -base YARanatr2+orig -source YARanatr1+orig -prefix ${ec}anatr1_1RegTo2 -verb -warp shift_rotate -cost mi -automask -1Dfile ${ec}anatr2toanatr1

• Average the two aligned T1s into one dataset

This takes approximately 30 seconds.

3dmerge -gnzmean -nscale -prefix ${ec}anatavg ${ec}anatr2+orig  ${ec}anatr1_1RegTo2+orig

Step 4: Make the surface model

• Create an 'fs' directory under the patient's directory to save the surface files from freesufer

/Volumes/data/scripts/@prep_dir {$ec}anatavg+orig.BRIK

• Create different cortical surfaces in freesurfer (i.e. inflated surface, pial surface, white matter surface etc.)

This takes several hours.

/Volumes/data/scripts/@recon {$ec}anatavg+orig.BRIK 

• Align the newly made freesurfer surfaces to the original experiment anatomy made by AFNI.

This creates the fs_SurfVol_Alnd_Exp file.

/Volumes/data/scripts/@finish {$ec}anatavg+orig.BRIK  

• Convert sulc files from freesurfer format to to ascii format that is readable by SUMA.

Sulc files contain information about the displacemnt of sulci and gyri from a hypothetical averaged surface.

cd ..
cd fs/surf
mris_convert -c lh.sulc lh.smoothwm lh.sulc.asc
mris_convert -c rh.sulc rh.smoothwm rh.sulc.asc
cp *sulc.asc ../SUMA

Now sulc files are converted to .asc format and copied into the SUMA folder.

• Make pial-outersmoothed surfaces in freesurfer

set subjid = fs
set SUBJECTS_DIR = /Volumes/data/UT/YAR
setenv SUBJECTS_DIR /Volumes/data/UT/YAR
set ec = fs
recon-all -s $ec -localGI

• Convert pial-outer_smoothed files from freesurfer format to ascii format

mris_convert lh.pial-outer-smoothed lh.pial-outer-smoothed.asc
mris_convert rh.pial-outer-smoothed rh.pial-outer-smoothed.asc

cp lh.pial-outer-smoothed.asc ../SUMA/ 
cp rh.pial-outer-smoothed.asc ../SUMA/

Now pial-outer_smoothed files are converted to .asc format and copied into the SUMA folder.

• Convert the freesurfer surface specification file into a format readable by SUMA

@SUMA_Make_Spec_FS -sid YAR

• Create skull stripped anatomy to allineate to the EPI

EPI images do not cover the skull. EPI and T1 match better if we crop the skull from the T1 images as well.

This takes approximately 3 minutes.

cd /Volumes/data/UT/YAR/afni
set ec = YAR
3dSkullStrip -input {$ec}anatavg+orig -prefix {$ec}anatavgSS  

• Transform the skull stripped anatomical image to match a template in TLRC space

This takes approximately 3 minutes.

@auto_tlrc -base TT_N27+tlrc -no_ss -input {$ec}anatavgSS+orig  

Step 5: Process CT data:

• Create 3D datasets for use with AFNI from these 2D CT image files

set ec = YAR
set session = /Volumes/data/UT/YAR/afni
cd /Volumes/data/UT/YAR/LastName_FirstName_CT/CTBRAINWITHOUTCONTRAST/Axial__Supine_Scan_3 
to3d -session $session -prefix {$ec}_CT IM-0001-0*

Step 6: Align the CT to the T1:

In this step use the 2nd T1 (YARanatr2+orig) as the base.

3dAllineate -base YARanatr2+orig -source {$ec}_CT+orig -cmass -prefix {$ec}CT_REGtoanat2 -verb -warp shift_rotate -cost mutualinfo -1Dfile {$ec}CT_REGtoanat2

- In afni display YARanatr2 as underlay and YARCT_REGtoanat2 as overlay to check if the alignment worked well.

Step 7: Locate the electrodes on the brain

• Tag the Electrodes

- Load AFNI and SUMA together

cd /Volumes/data/UT/YAR
set ec = YAR
./@ec

- In afni set underlay to fs_SurfVol_Alnd_Exp+orig and overlay to YARCT_REGtoanat2

Setting the background to overlay to see just see the CT scan sometimes makes it easier to see the electrodes. 'u' key can be used to switch the background between underlay and overlay.

- Go to Define Datamode --> Plugins --> Edit Tagset

- Set Dataset as fs_SurfVol_Alnd_Exp

- Name Tag File as YARelectrodetags.tag

- Find the densest view of an electrode, make sure it's centered in all views (axial, saggital and coronal) and press 'Set' and then 'Save'

- Press 'Write' once all the electrodes are tagged. This will write the YARelectrodetags.tag file and save it in the afni folder.

- Open the tag file YARelectrodes.tag in Excel and re-order the electrode tags in the order of the channel numbers

- Make a copy of YARelectrodes.tag with the file name extension .1D

cp {$ec}electrodetags.tag {$ec}electrodetags.1D

- Open in {$ec}electrodetags.1D in Excel and edit the file so that it only has the 3 columns of x,y,z

- Check fs_both.spec file in fs/SUMA, make sure it contains surface specifications for the pial-outer-smoothed surfaces. Because tagged electrodes will be projected to onto the pial-outer-smoothed surfaces. If fs_both.spec file does not contain surface specifications for the pial-outer-smoothed surfaces, then copy these lines into the end of fs_both.spec file:

NewSurface
       SurfaceFormat = ASCII
       SurfaceType = FreeSurfer
       FreeSurferSurface = rh.pial-outer-smoothed.asc
       LocalDomainParent = rh.smoothwm.asc
       SurfaceState = pial-outer-smoothed
       EmbedDimension = 3
       Anatomical = N

NewSurface
       SurfaceFormat = ASCII
       SurfaceType = FreeSurfer
       FreeSurferSurface = lh.pial-outer-smoothed.asc
       LocalDomainParent = lh.smoothwm.asc
       SurfaceState = pial-outer-smoothed
       EmbedDimension = 3
       Anatomical = N

• Project the tagged electrodes onto the closest node on the pial-outer-smoothed surface

SurfaceMetrics -spec /Volumes/data/UT/YAR/fs/SUMA/fs_both.spec -sv fs_SurfVol_Alnd_Exp+orig -surf_A lh.pial-outer-smoothed.asc -surf_B rh.pial-outer-smoothed.asc 
-closest_node YARelectrodetags.1D -prefix {$ec}ClosestNodes_pialenvelope

- Open YARClosestNodes_pialenvelope.1D.dset in Excel and copy 3 columns of Xn, Yn, Zn coordinates. Paste Xn, Yn, Zn coordinates in a new Excel sheet and enter #spheres in the first cell of the document and format it as follows:

Xn Yn Zn 1 1 1 1 1.5 2.2

Electrode color is black for [0 0 0] and white for [1 1 1]. Electrode size is 1.5 for regular, 0.5 for mini electrodes.

- Save it as YARClosestProjected.txt and copy is as a .do file

cp {$ec}ClosestProjected.txt {$ec}ClosestProjected.do

• Display the projected electrodes on the brain surface

- Open "Select Displayable Objects File" shell by pressing "Comman & Control & s" on the keyboard. Set filter in shell to *do and select the appropriate .do file.

HINTs to make electrode tagging easier:

1) Display electrodes on the brain surfer in SUMA window

- Load AFNI and SUMA together by the ./@ec command

- Set underlay to fs_SurfVol_Alnd_Exp+orig and overlay to YARCT_REGtoanat2

- Increase the threshold so that in the SUMA window, electrodes look as colorful spots on the gray brain surface. Right clicking on these colorful spots on the SUMA window will take the Xhairs to the corresponding electrode position in the AFNI display window.

2) Display electrodes on the brain by using AFNI C Renderer

- Go to Define Datamode --> Plugins --> Render Dataset

- Choose YARanatSS as underlay and YARCT_REGtoanat2 as overlay

- Select See Overlay and DynaDraw. Increase the threshold so that everthing on the brain surface doesn't look green. Hit automate to generate the surface image with electrodes on it

- Play with the Roll, Pitch and Raw values to see the brain surface from different views

- Select See Xhairs. This way the Xhairs in the rendering window will move together with the Xhairs in the AFNIdisplay window

Step 8: Label the electrodes

- Go to Define Datamode --> Plugins --> Edit Tagset

- Load the electrode tags with the same dataset setting (fs_SurfVol_Alnd_Exp)

- Enter a Tag Label for each electrode and press 'Relabel' and 'Save'

- Press 'Write' after relabeling all the electrodes

• In Matlab, generate a generateOBJSetting.m script for the subject with the channel list.

• Generate a YAR_Label_Input.txt file with the closest projected x,y,x coordinates and run the read_write_elec_labels.m script.

• Copy the generated YAR_ElectrodeLabels.niml.do file into the patient's afni folder and paste the following line at the beginning of the file:

<nido_head													
coord_type	=	mobile											
default_SO_label	=	CURRENT		
bond	=	none	
/>	

Electrode labels can now be displayed on the brain surface next to the corresponding electrodes. In SUMA window, "Select Displayable Objects File" shell by pressing "Command & Control & s" on the keyboard. Set filter in shell to *do and select the appropriate .do files that correspond to the electrode locations (i.e. YARClosestProjected.do) and electrode labels (i.e. YAR_ElectrodeLabels.niml.do).