DICOM orientation

as_dub · Post by **as_dub** » Thu Oct 20, 2005 9:14 pm

Hello !

I have DICOM anatomical MRIs from which I want to extract the spherical meshes. After importing them with "import T1 MRI" function, I see in Anatomist that the orientation is wrong (sagittal image appear in the axial windows, coronal in the sagittal and axial in the coronal)
When I run the "Ana do a lot of things" without changing the "allow_flip_initial_MRI" (false) of the "prepare subject" function I obtain spherical meshes but in the wrong orientation.
Whereas If I switch the "allow_flip_initial_MRI" to true, I observe that the coronal axial and sagittal views appear in the right windows (in Anatomist), BUT the "Ana do a lot of things" crash.
Could you help me to solve this problem please ? I have 3 sets of acquisition DATA (same acquisition machine and same protocol), and the problem is not the same for each set... If it can help I can send you the file where I reported log of errors for each set, but if you have any suggestion to solve the problem I could try it before. Thanks !

AnneSo

Post by **riviere** » Fri Oct 21, 2005 10:20 pm

Hi,

The initial orientation of images (just after importation) depends on how the acquisition was done, so it can be normal that axial windows display sagittal imahes.
The PrepareSubject process is supposed to flip images (if allowed by the flag) so they are in the right orientation.
Normally the T1 segmentation pipeline fails if images are not in the expected orientation. What you get here is exactly the contrary, so there is obviously a problem somewhere.

Can you check the AC/PC/IP positions after flipping (using the viewer) please ? If they have not correctly been updated when the image was rotated, it could explain failures.
Otherwise, yes please, send the logfile to the support address.

Denis

Jean-Francois Mangin · Sat Oct 29, 2005 3:10 pm

Do you have some trouble with all the Dicom images you get, or only with a couple?

as_dub · Post by **as_dub** » Mon Oct 31, 2005 7:45 pm

Almost all. For now I have 4 datasets, and only one is good giving results..

denghien · Post by **denghien** » Wed Nov 02, 2005 9:49 am

Hi,

If you want I can test your data. So, give me your mail.

Isa

as_dub · Post by **as_dub** » Tue Dec 06, 2005 10:16 pm

Hello Isa,
if you still have a little time to test my data, it would be great ! here is my email adress as_dub@hotmail.com. Send me an email !! Thank you,
Anne-Sophie