open-discussion > Differing electrode coordinates (BIS/ FSL)
Showing 1-5 of 5 posts
Display:
Results per page:
Mar 7, 2017  11:03 PM | Johann Philipp Zöllner - Goethe University Frankfurt
Differing electrode coordinates (BIS/ FSL)
Dear all,

I need to extract the positions of depth electrodes from CT images (in subject space). I currently coregister post-implant CT to pre-implant MRI using FSL/Flirt which works fine. However, when I open the coregistered CT in BioImageSuite, the electrode positions - when defined and exported using the electrode editor - seem to be off/wrong, the CT itself seems to be shifted as well as flipped along the X-axis compared to the coordinates I get in fslview (or 3D Slicer for that matter, which gives exactly the same positions as fslview). I assume this has to do with differing interpretation of the header information among the programs?
Can anyone tell me how to set up BioImageSuite to get corresponding electrode localisations in subject space? I find the rest of the editor very helpful, so any suggestions would be highly welcome.

Thanks in advance,
Johann Philipp Zöllner
Mar 8, 2017  07:03 AM | Xenophon Papademetris
Differing electrode coordinates (BIS/ FSL)
BioImage Suite automatically maps your images to LPS "on load" so this accounts for the flipping.

One solution that *may* work is to change this default.

Under Help|Preferences go to File Formats and set "ForceBISOrientationsOnLoad" to Off.

Then restart the program and just try a simple couple of electrodes and see if this fixes things.

[ I will be on the road the next few days so I may not have access to reply.]
Mar 8, 2017  12:03 PM | Johann Philipp Zöllner - Goethe University Frankfurt
RE: Differing electrode coordinates (BIS/ FSL)
Dear Mr. Papademetris,

thanks for your quick reply. I was able to partially solve the problem by just re-orienting the image to the RAS orientation in the image editor. Unfortunately, I could not find the variable "forceBISorientation" under Preferences->file format, but manually re-orienting should be fine.

However, when I try to threshold the image after re-orienting, I get a VTK error message (attached)

Thanks again,
Johann Philipp Zöllner
Attachment: error_vtk_BIS.jpg
Mar 8, 2017  01:03 PM | Xenophon Papademetris
RE: Differing electrode coordinates (BIS/ FSL)
Try making the preferences dialog bigger and see if it shows up.

If this does not work: Save the CT from BioImage Suite out and then use this version (i.e. the saved one) for registration. FSL may work better with this as, at least, it will be consistent with the electrodes. Otherwise send me a direct e-mail early next week (Xenophon dot Papademetris at yale.edu) and we can figure it out.
Mar 8, 2017  02:03 PM | Larry Olson - Emory University
RE: Differing electrode coordinates (BIS/ FSL)
> I have always used BIS to coregister CT/electrodes to (unstripped) MRI. It
> works great. It will solve your problem. You can threshold the CT to
> extract the electrodes as outlined in the manual, and touch up with editing
> any trouble spots in another program (I use MRIcron).
>

The only problem with CT and BIS is that an extremely large file will crash
the program (maybe memory allocation issue?), in which case you may need to
scale the CT down first.

I think your trouble could lie in the file format you use. FSL likes to
save in nifty, and BIS in Analyze preferentially, and the 2 programs may
interpret the headers differently.

If I do create a mismatch in orientations for some reason (the images are
still coregistered), I find it easier/more intuitive to "correct" that in
Amide than in BIS. Amide has pretty simple rotations (just invert the axis
as needed), but an arcane interface.

If you have 2 volumetric CTs, one with electrodes and one without, a nice
method to extract electrodes is to "subtract" the one without from the one
with.