Hi Sooyun,
It is possible that tri-linear might work better, but I have to caution that you will get non-integer values. Have you tried mri_label2vol? According to Doug in the last email, mri_label2vol might work better than mri_vol2vol.
My suggestion is to first verify whether the transformation from Choi parcellation to the subject's native space is good enough. To do that, instead of transforming the Choi parcellation, try transforming the MNI152 template accompanying the Choi's parcellation (NOT the FSL MNI152 template). If the transformed MNI152 template does not match your subject's native anatomy to your liking, then you need to improve the transformations.
Again, can you clarify where you are obtaining the transformation matrices? Recall that I previously told you that to convert the Choi's parcellation to the MNI152 template from FSL, you should just use mri_vol2vol, rather than FIRST. The command is as follows: "mri_vol2vol --mov original_choi_MNI152_parcellation.nii.gz --targ $FSLDIR/data/standard/MNI152_T1_2mm_brain.nii.gz --regheader --o FSL_choi_MNI152_parcellation.nii.gz --no-save-reg --interp nearest". And the reason is that the Choi volume is in exactly the SAME space as the $FSLDIR/data/standard/MNI152_T1_2mm_brain.nii.gz, it's just that the header information is different. Therefore the "regheader" flag takes care of that.
Cheers, Thomas
On Tue, Feb 11, 2014 at 2:03 PM, 조수연 syduam2452@kaist.ac.kr wrote:
Hi, Thomas.
I have circled the unmasked regions in the image with a red brush tool. The yellow-red regions are the FreeSurfer space->MNI152 space->Subject space transformed Choi's loose mask files, and the blue regions are the subject's subcortical areas. My current ROI is the caudate and the putamen. As you can see, there are multiple regions where the labels aren't big enough to cover my subject data. I checked the original Choi's parcellations in the original space and the MNI152 space but it's hard to check whether the label in the specific location of my interest is missing or not because the voxel numbers and dementions are different per space, thus is hard to get the specific coordinates that lacks mask.
I discussed with my lab members and concluded maybe the problem lies when we are transferring the image from one space to another in that we used the nearest neighbour when applying the transformation matrix. We are planning to use tri-lear for interpolation instead so that the achieved image has a looser intensity specificaton, so that it helps us to specify the lables of the subject's regions which are not covered by Choi's masks. Hope this helps.
If you have any suggestions as of how to morph Choi's images best to our subject's fit, please do inform me. Thanks!
Best regards, Sooyun.
-----------------------원본메일----------------------- 보낸사람 : Thomas Yeo ythomas@csail.mit.edu 받는사람: 조수연syduam2452@kaist.ac.kr 참 조: freesurfer freesurfer@nmr.mgh.harvard.edu 보낸날짜: 2014-02-11 00:11:48 GMT +0900 (ROK) 제목: Re: [Freesurfer] Choi's parcellation expansion inquiry
Hi Sooyun,
Which is the sky blue region you are referring to? Can you place your fslview cursor in the location where you think is missing the label?
Can you visually inspect the original Choi's parcellation to see if the label is also missing at the corresponding MNI152 location? If the label is not missing, then perhaps your transformation is not doing its job, which I suspect might be the problem since the loose mask in Choi's parcellation is already pretty loose.
Regards, Thomas
On Mon, Feb 10, 2014 at 12:20 PM, 조수연 syduam2452@kaist.ac.kr wrote:
Hi, all.
I'm trying to apply Choi's parcellations onto my subjects' data. I used the 7-Network loose mask but it turns out that even the loose mask isn't capable of covering up the ROI.
Is there a way I can further "expand" the loose mask and make it "looser"?
The uncovered areas of my subject's image are colored skyblue. The rest are parcellated segments of Choi's image.
FYI Choi's FreeSurfer grid images were linearly transformed by concat matrix which shifts FreeSurfer grid to Standard MNI152 grid to our subject's grid.
Thanks.
Best regards, Sooyun.
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at http://www.partners.org/complianceline . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail.