OpenMRSLab and voxel visualization


I am trying to reproduce voxel visualization for GE SVS in OpenMRSLab. I am a bit confused due to some space rotations I can’t handle in my mind. It looks like GE anatomical scan is stored in sagittal in contrast to Siemens T1, which is an axial one in the tutorial. The same for the voxel coordinates… I am not sure about the orientation.
Dear Ben, please give me some support. I attach my notebook but I would appreciate some other examples as well.
Finally, I would like to ask the community to share Your thoughts about voxel-to-image conversion. Do You know some tools, which could simply create a mask? For Siemens I used my own code released on NITRC, but it became more complicated when we have to deal with also GE etc. data…

Looking forward to hearing from You
Bartosz Kossowski

My file is here ge_voxel.pdf (264.4 KB)

Hi Bartosz,

Just enabled PDF attachments - sorry, the forum software has pretty restrictive default settings for this.

Regarding your voxel mask question - both Gannet and Osprey have built-in support for co-registering of GE data.


Hi Bartosz,
I have just pushed a new notebook to the suspect.readthedocs page (here), the pdf version is attached here too. Hopefully that will be more complete and help you get things working. Thanks for sharing the notebook, that is very useful and thorough. I think the problem that you are having is getting the coordinates for plotting the voxel edges - although the image is sagittal, by using the from_scanner() function you move into the coordinate system of the image volume, so you should still use coords[0] and coords[1] to get the column and row positions of the voxel. It is only if you are pulling e.g. a sagittal slice out of an axially ordered volume that you need to take a different pair out of the (x,y,z) tuple. In the new notebook (which uses a GE dataset) I show how to plot the voxel outline on slices in all three planes, and how to get a voxel mask as well. Obviously those two ways of rendering the voxel should line up with each other.

If you could share the P-file and FSPGR with me, and the reference images, I am happy to check and make sure things are working. Also if you have any suggestions for how to improve the guide notebook, I would be keen to hear them, I just wrote it today and it could do with some peer review.

Co-registering Images.pdf (297.4 KB)

Thank You both for the quick reply. I will try both “Gannet CoRegStandAlone” function and the new OpenMRSLab tutorial!