help > Inconsistent Proportions
Showing 1-5 of 5 posts
Nov 2, 2016 05:11 PM | Aaron Brenningmeyer - Southern Illinois University
Inconsistent Proportions
I recently fixed a segmentation
error and the code is still attached to this forum, however I have
ran into a new problem. The yuki program has errors when reading
the frontal region occasionally due to a artery that runs near the
Corpus Callosum. The program often registers this artery as part of
the Corpus Callosum and gives inaccurate readings of the image. Is
there a way to prevent this?
Nov 14, 2016 05:11 PM | Babak Ardekani
RE: Inconsistent Proportions
Aaron,
I have seen this problem before whereby the signal in the pericallosal artery is enhanced due to inflow artifacts (flow of fully magnetized protons into the imaging slice) which makes the artery indistinguishable from the corpus callosum. This appears in some pulse sequences and not others, which leads me to believe that the problem can be prevented by using a well-designed 3D T1W sequence. Unfortunately, there is not much that can be done after the fact other than manually correcting Yuki's output.
Babak
I have seen this problem before whereby the signal in the pericallosal artery is enhanced due to inflow artifacts (flow of fully magnetized protons into the imaging slice) which makes the artery indistinguishable from the corpus callosum. This appears in some pulse sequences and not others, which leads me to believe that the problem can be prevented by using a well-designed 3D T1W sequence. Unfortunately, there is not much that can be done after the fact other than manually correcting Yuki's output.
Babak
Nov 30, 2016 05:11 PM | Aaron Brenningmeyer - Southern Illinois University
RE: Inconsistent Proportions
Babak,
Thank you for your feedback! Is there a way that we could manually edit the output of Yuki? On the image there are multiple + symbols. Is there a way that we could adjust those to appropriately refit the margins to exclude the pericallosal artery? I provided an image to show an example of the problem that we have occuring.
Thank you,
Aaron
Thank you for your feedback! Is there a way that we could manually edit the output of Yuki? On the image there are multiple + symbols. Is there a way that we could adjust those to appropriately refit the margins to exclude the pericallosal artery? I provided an image to show an example of the problem that we have occuring.
Thank you,
Aaron
Dec 1, 2016 07:12 PM | Muhammad Asim Mubeen - Olin Neuropsychiatry Research Center, Hartford Hospital
RE: Inconsistent Proportions
Aaron,
In this case you have to manually draw corpus callosum (CC) and then run the yuki again. Here are the steps.
Regards,
--
Asim
In this case you have to manually draw corpus callosum (CC) and then run the yuki again. Here are the steps.
-
Open the MSP file (in your case it may be named as cos002a1001_msp.nii) in an image-editing tool (ITKSNAP or FSLVIEW), where you can draw and save a segmentation. My preference is ITKSNAP.
-
Draw the CC manually on the the MSP image and save the segmentation as a file cos002a1001_cc_corrected.nii (This corrected image should be a binary image)
-
Now you can run the yuki with -cc option as follows:
-
yuki -ppm -cc cos002a1001_cc_corrected.nii -i cos002a1001.nii -csv cos002a1001_corrected.csv -H
-
The important option here is the -cc option that you should use to mention the corrected binary CC image.
Regards,
--
Asim
Dec 2, 2016 01:12 PM | Babak Ardekani
RE: Inconsistent Proportions
Thanks for the reply Asim. I would add that in Step 2 you
won't have to draw the CC manually from scratch. You can load
the imperfect cos002a1001_cc.nii generated by yuki into ITKSNAP as
a segmentation image and mend it using brush tools and after you're
happy with it save it as cos002a1001_cc_corrected.nii.
Babak
Babak