I've attached a screenshot of an error I am having while localizing electrode grids. The grids are localized properly, but are not aligned with the surface of the brain after I strip the skull (in image processing). 

thanks for any ideas!

Can't see the attachment. You may want to upload the picture somewhere and send a link to it. Having said this, it would be more helpful if you could outline the procedure you use. My hunch is that the electrodes are still in CT-space and need to be transformed.

I'll try re-sending the screenshot. It's as if the grid is hovering above the surface of the brain, rather than on it. How do I transform out of CT space?

the most accurate placement I've been able to obtain is by transforming the grid with the inverted matrix produced from linearly registering MRI to CT, however it is far from exact (see attached). any other ideas would be much appreciated.

Hi Hanna,

I have been using the electrode editor for a while and it works pretty good. Eventually I got offset but it is due I was co registering the CT with a pre-op MR. If you co-register the CT with a post-op MR the result is very accurate.

In your case, did you use a pre or post- op MR? Did you double check the coregistered MR? Once I use the MR co-registered, the electrodes fall in position. Eventually I do a manual adjustment. But in general it works pretty well.

Based on your image it looks the MR space is different from the CT space where the electrode objects were generated.

Cheers

Magno

The MR I'm using is pre-op, but I've co-registered to CT. I'll try using post-op, and double-checking the registration. thanks!

You are probably looking at the effect of brain deformation as a result of putting stuff in the brain.

The best procedure is:

Nonrigid mapping between preop MRI and postop MRI (we do this using stripped MRIs)
Rigid mapping between postop MRI and postop CT (use unstripped images for this one)

Then invert both transformation (the nonrigid one .grd will take a while) and concatenate them in the ElectrodeEditor transform Grid tool to move the electrodes back.

Xenios

I am using the electrode editor for several months without problems now. But I remember similar problems in the beginning. Usually I did localize the electrodes in the original CT, not the coregistered one. My recipe to visualize electrode grids is the following:

1. Make sure you coregister the low resolution image (CT) with the high resolution image (MR). The high resolution image should not be changed.

2. Make sure the original CT has a different name than the coregistered CT. I do append "_coreg" to the name of the coregistered CT.

3. Localize the electrodes in the coregistered CT (i.e. *_coreg). This is the most important step!

4. Extract the brain from the original MR.

5. Load the extracted brain from the original MR into the electrode editor.

Now the position of the electrodes should match the correct positions very well if (!) the original MR was a post-OP image. Because in this case the effect of a subdural hematoma (a locally "squeezed" brain) will already be taken into account. In a pre-OP MR image the electrodes would appear a bit below the surface.

Just a last remark: I use FSL/flirt for step 1 and FSL/bet for step 4.

I hope this helps,

Markus

this is the process I have used as well, although with pre-op MRI and post-op CT. Would post-op MRI reduce the difference in localization enough?

Yes. Doing a nonlinear registration between the preop MRI to the post of MRI and then a linear mapping from post op MRI to post op CT will improve things as it will account for the deformation of the brain caused by the implantation.

Do you have any guidance on modifying the ElectrodeExportAsText module to customize the output to be [Electrodename][electrodenumber] delimiter [xcoor] delimterer [ycoor] delimeter [zcoor] so it can be used in other matlab programs (ie EEGLAB or brainstorm)?
 
Andrew

Andrew - sorry for the delay - I am on the road. This is doable. If you sent an example text file we should be able to modify the output (once I get back to the office). [ Or more likely write a quick script to reformat the existing output].

Xenios

Thank you Sir.  Just tell me when and where to email it and I will.  I really appreciate it.
 
I attached the matlab version of an electrode file, but it will be fine if it is in text format like above.

Xenophon dot Papademetris at yale dot edu