Dog brain DTI problems

(Ors Petnehazy) #1

Dear Users,

My name is Ors Petnehazy and I’m interested in 3D veterinary anatomical reconstrucions. I know some features of the Slicer. I made some reconstructions already (air sacs, horse stifle CT, MR, etc.).
For a research project I need to reconstruct a dog brain DTI series. I followed the instructions I found made by Sonja Pujol with my own dataset. After some step I only get a dark “image” as a result. I’d like to ask for your help for DTI reconstruction of a dog brain DTI dataset.


Ors Petnehazy

(Isaiah Norton) #2

Hi Ors,

Welcome to the forum. Could you clarify your workflow please?

  • What Slicer version? What operating system?
  • What scanner did you use to scan?
  • Do you have DICOM images, and are you converting with DWIConvert? or some other steps?
  • Which step is run before you get the dark image?

It may also be helpful if you can share the data, either converted or raw DICOM, using dropbox/onedrive/etc. If you prefer to share privately please email the download link to me at inorton -at- bwh dot harvard dot edu. (hopefully there should not be any confidentiality issues :dog:).

(Ors Petnehazy) #3

I followed the steps again, written in the DiffusionMRIanalysis PDF. I attach the screen savings I made during the procedure.
First, I’m not familiar with the DTI procedure, I understood only the basics but I try to do my best to improve in it.
The dataset comes from a 3T Biograph (Siemens). I tried to find any information about the resulting files of the DTI measurement but I was not sure what dataset I have to open for the reconstrucions. I tried all of these for doing the DICOM to NRRD transformation but there was only one where the transformation was successfull. So I worked with this dataset.
It was interesting that after each step when I changed the view from trans (red) to sagittal (yellow) for example the blue brain mask was on the new orientation. After making the brain mask, I saw that it contains the temporal muscles and the whole head as well… I was sure after this stpe that there is something not working properly…

I send you the dataset and the screen captures via wetransfer.

Kind regards,


(Ors Petnehazy) #4

Dear Isaiah,

Thank you for your offer. I wrote my impressions in a separate message.



(Isaiah Norton) #5

@opetne thanks for sharing the data. The first dataset ep2d_diff_mddw_64_p2 seems quite distorted and I didn’t get good results from it.

The second dataset ep2d_diff_mddw_64_p2_4100_tra looks pretty good. I recognize the major structures anyway (I don’t know anything about canine neuroanatomy!). I followed these steps:

  • DWIConvert
  • Diffusion Brain Masking
  • Diffusion Tensor Estimation
  • Tractography Seeding module (see the “Slicer Neurosurgical Planning Tutorial” linked here).

From your screenshots, I think you did the same except the last step.

The resulting tracts seem ok (again with no specific knowledge). Here is a screenshot:


It was interesting that after each step when I changed the view from trans (red) to sagittal (yellow) for example the blue brain mask was on the new orientation.

I think you may have several datasets open? They have different image sizes, so it could lead to some confusion. Also, you should use the DWI volume ep2d_diff_mddw_64_p2_4100_tra in Slicer. The other ones with e.g. _TR or _FA in the name are calculated on the scanner, and there seems to be some change in the orientation/origin so the images are not directly comparable.

Your screenshots look good so far, except the mask overlay issue. So, I suggest to work with your dataset ep2d_diff_mddw_64_p2_4100_tra, and also try some of the example tutorial data from the link above to learn more processing steps. If you have any more questions, please let us know.


(Ors Petnehazy) #6

Dear Isaiah,

Thank you so much for this explanation. You are right the images seem
distorted for me too, at the beginning I was wondering if there will be any
useable results from this dataset.
For my trials i was using only one dataset, I have to check which one it
was but I guess it could comprise more image orientation.
I’ll go trough the steps again and hopefully I can have better results.



Isaiah Norton (időpont: 2018. jan. 19., P,
18:39) ezt írta: