Hello,
I am trying to validate the ICV value by lining up the nu_noneck image with the RB image. These are the commands I am using:
mri_convert nu_noneck.mgz --apply_transform transforms/talairach_with_skull.lta nu_noneck5.mgz tkmedit -f nu_noneck5.mgz & tkmedit -f /apps/freesurfer/average/RB_all_withskull_2006-02-15.gca &
When I compare these two, they don't line up. The RB image is all blurry (Maybe it's a combination of several subjects and is supposed to be like that). Any suggestions as to how to line up the noneck transformed imaged to the RB image is greatly appreciated!
I also attached three files to show what's happening in my transform process.
Thanks, Jeff
----------------------------------------
Date: Fri, 5 Sep 2008 20:31:55 -0400 From: fischl@nmr.mgh.harvard.edu To: jsadino@hotmail.com Subject: RE: [Freesurfer] Talairach_with_skull.lta accuracy
can you send this to the list so others can respond? Where did you get the tmp.reg file? If you have transformed them already you shouldn't need tkregister2 (or a registration file in general) to compare them.
On Fri, 5 Sep 2008, Jeff Sadino wrote:
Thank you very much for your help. Your knowledge and quick responses are very much appreciated.
I converted my image into transformed space, but it still does not match the RB image. These are the commands I used:
mri_convert nu_noneck.mgz --apply_transform
transforms/talairach_with_skull.lta nu_noneckt3.mgz
tkregister2 --targ
/apps/freesurfer/average/RB_all_withskull_2006-02-15.gca --reg tmp.reg --lta /data/040002_S04_bakj/mri/transforms/talairach_with_skull.lta --mov /data/040002_S04_bakj/mri/nu_noneckt3.mgz
Again, my main goal is to verify the ICV value, so to do that I need to verify that the lta file is transforming the nu_noneck image onto the RB image...I think.
What step am I missing?
Thank you Jeff
Date: Fri, 5 Sep 2008 13:09:21 -0400 From: fischl@nmr.mgh.harvard.edu To: jsadino@hotmail.com CC: greve@nmr.mgh.harvard.edu Subject: RE: [Freesurfer] Talairach_with_skull.lta accuracy
that won't generate a transformed volume, just display it in the transformed space. That is, the volumes you sent were in their native space and would have no reason to be aligned. The neck removal seems to have been fine in both cases. On Thu, 4 Sep 2008, Jeff Sadino wrote:
ps, sorry, I forgot to include my command:
tkregister2 --targ /apps/freesurfer/average/RB_all_withskull_2006-02-15.gca --mov /data/${file}/mri/nu_noneck.mgz --lta /data/${file}/mri/transforms/talairach_with_skull.lta --reg tmp.reg
I believe this will transform the nu_neck.mgz image via the talairach matrix, so that it matches the RB image.
btw, we processed the scans using freesurfer 3, but the tkregister2 that came with freesurfer 3 did not include the --lta tag, so I am using the tkregister2 from freesurfer4.
Thanks!
Date: Thu, 4 Sep 2008 21:49:27 -0400 From: fischl@nmr.mgh.harvard.edu To: jsadino@hotmail.com CC: freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] Talairach_with_skull.lta accuracy
Hi Jeff,
why do you say that one is bad? We don't resample, so neither one will be aligned with the atlas unless you apply the transform.
We are aware that the eTIV measures have more variability than we would like, and we're looking into it, but don't have a good solution from only T1 images.
cheers, Bruce On Thu, 4 Sep 2008, Jeff Sadino wrote:
Hello all,
According to aseg.stats, our ICV values fluctuate on the same subject across scans, sometimes by up to 10% I have been validating the talairach_with_skull.lta accuracy by comparing the nu_neck.mgz to the RB_all_withskull_2006-02-15.gca template. I read on the boards that it is not all that necessary that the brains match, that it is more important for the skulls to match. The RB is blurred (I don't fully understand where the RB comes from, I think from 22 subjects that the freesurfers combined into one?), so am I matching up the whitest areas of the RB to the skull on the nu_neck image? I noticed also that the bottom of the brains rarely match at all to the RB image...is this normal? Are there any other parts of the image that need to be checked for validity?
I attached two nu_noneck.mgz files, one was good and the other bad. Are these images typical?
Thank you for your help, Jeff
See how Windows Mobile brings your life together?at home, work, or on the go. http://clk.atdmt.com/MRT/go/msnnkwxp1020093182mrt/direct/01/
Get more out of the Web. Learn 10 hidden secrets of Windows Live. http://windowslive.com/connect/post/jamiethomson.spaces.live.com-Blog-cns!55...
See how Windows connects the people, information, and fun that are part of your life. http://clk.atdmt.com/MRT/go/msnnkwxp1020093175mrt/direct/01/
_________________________________________________________________ See how Windows Mobile brings your life together—at home, work, or on the go. http://clk.atdmt.com/MRT/go/msnnkwxp1020093182mrt/direct/01/
At some point in early 2008, Doug had created a version of tkregister2 that he graciously provided us, which had --gca-skull and --gca flags, which greatly simplified the process of confirming the accuracy of either the talairach_with_skull.lta and talairach.lta transforms (respectively). However, neither of those flags are options in the current public release of FS 4.0.5.
Doug: Will the forthcoming FS release have those flags available as options to tkregister2? We've found them to be a very helpful addition to tkregister2, and it sounds like Jeff would as well.
Jeff: In the interim, here are my notes on how to do what you want to accomplish (N.B. Doug's improved tkregister2 using the --gca-skull flag is a more powerful approach because it makes the target the subject space and thus the surfaces can also be meaningfully displayed via the --surf option):
mri_convert $FREESURFER_HOME/average/RB_all_withskull_2007-08-08.gca RB_gca.mgz --frame 0 (N.B. add --conform option if you want to avoid format conversion within tkregister2 itself (which slows loading))
tkregister2 --targ RB_gca.mgz --mov nu_noneck.mgz \ --lta transforms/talairach_with_skull.lta --reg tmp.reg
cheers, -Mike H.
On Mon, 2008-09-08 at 15:54 -0600, Jeff Sadino wrote:
Hello,
I am trying to validate the ICV value by lining up the nu_noneck image with the RB image. These are the commands I am using:
mri_convert nu_noneck.mgz --apply_transform transforms/talairach_with_skull.lta nu_noneck5.mgz tkmedit -f nu_noneck5.mgz & tkmedit -f /apps/freesurfer/average/RB_all_withskull_2006-02-15.gca &
When I compare these two, they don't line up. The RB image is all blurry (Maybe it's a combination of several subjects and is supposed to be like that). Any suggestions as to how to line up the noneck transformed imaged to the RB image is greatly appreciated!
I also attached three files to show what's happening in my transform process.
Thanks, Jeff
Date: Fri, 5 Sep 2008 20:31:55 -0400 From: fischl@nmr.mgh.harvard.edu To: jsadino@hotmail.com Subject: RE: [Freesurfer] Talairach_with_skull.lta accuracy
can you send this to the list so others can respond? Where did you get the tmp.reg file? If you have transformed them already you shouldn't need tkregister2 (or a registration file in general) to compare them.
On Fri, 5 Sep 2008, Jeff Sadino wrote:
Thank you very much for your help. Your knowledge and quick responses are very much appreciated.
I converted my image into transformed space, but it still does not match the RB image. These are the commands I used:
mri_convert nu_noneck.mgz --apply_transform
transforms/talairach_with_skull.lta nu_noneckt3.mgz
tkregister2 --targ
/apps/freesurfer/average/RB_all_withskull_2006-02-15.gca --reg tmp.reg --lta /data/040002_S04_bakj/mri/transforms/talairach_with_skull.lta --mov /data/040002_S04_bakj/mri/nu_noneckt3.mgz
Again, my main goal is to verify the ICV value, so to do that I need to verify that the lta file is transforming the nu_noneck image onto the RB image...I think.
What step am I missing?
Thank you Jeff
Date: Fri, 5 Sep 2008 13:09:21 -0400 From: fischl@nmr.mgh.harvard.edu To: jsadino@hotmail.com CC: greve@nmr.mgh.harvard.edu Subject: RE: [Freesurfer] Talairach_with_skull.lta accuracy
that won't generate a transformed volume, just display it in the transformed space. That is, the volumes you sent were in their native space and would have no reason to be aligned. The neck removal seems to have been fine in both cases. On Thu, 4 Sep 2008, Jeff Sadino wrote:
ps, sorry, I forgot to include my command:
tkregister2 --targ /apps/freesurfer/average/RB_all_withskull_2006-02-15.gca --mov /data/${file}/mri/nu_noneck.mgz --lta /data/${file}/mri/transforms/talairach_with_skull.lta --reg tmp.reg
I believe this will transform the nu_neck.mgz image via the talairach matrix, so that it matches the RB image.
btw, we processed the scans using freesurfer 3, but the tkregister2 that came with freesurfer 3 did not include the --lta tag, so I am using the tkregister2 from freesurfer4.
Thanks!
Date: Thu, 4 Sep 2008 21:49:27 -0400 From: fischl@nmr.mgh.harvard.edu To: jsadino@hotmail.com CC: freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] Talairach_with_skull.lta accuracy
Hi Jeff,
why do you say that one is bad? We don't resample, so neither one will be aligned with the atlas unless you apply the transform.
We are aware that the eTIV measures have more variability than we would like, and we're looking into it, but don't have a good solution from only T1 images.
cheers, Bruce On Thu, 4 Sep 2008, Jeff Sadino wrote:
> > Hello all, > > According to aseg.stats, our ICV values fluctuate on the same subject across scans, sometimes by up to 10% I have been validating the talairach_with_skull.lta accuracy by comparing the nu_neck.mgz to the RB_all_withskull_2006-02-15.gca template. I read on the boards that it is not all that necessary that the brains match, that it is more important for the skulls to match. The RB is blurred (I don't fully understand where the RB comes from, I think from 22 subjects that the freesurfers combined into one?), so am I matching up the whitest areas of the RB to the skull on the nu_neck image? I noticed also that the bottom of the brains rarely match at all to the RB image...is this normal? Are there any other parts of the image that need to be checked for validity? > > I attached two nu_noneck.mgz files, one was good and the other bad. Are these images typical? > > Thank you for your help, > Jeff > > _________________________________________________________________ > See how Windows Mobile brings your life together?at home, work, or on the go. > http://clk.atdmt.com/MRT/go/msnnkwxp1020093182mrt/direct/01/
Get more out of the Web. Learn 10 hidden secrets of Windows Live. http://windowslive.com/connect/post/jamiethomson.spaces.live.com-Blog-cns!55...
See how Windows connects the people, information, and fun that are part of your life. http://clk.atdmt.com/MRT/go/msnnkwxp1020093175mrt/direct/01/
See how Windows Mobile brings your life together—at home, work, or on the go. http://clk.atdmt.com/MRT/go/msnnkwxp1020093182mrt/direct/01/ _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
I just added those changes to stable, so it should be in our next stable release.
Michael Harms wrote:
At some point in early 2008, Doug had created a version of tkregister2 that he graciously provided us, which had --gca-skull and --gca flags, which greatly simplified the process of confirming the accuracy of either the talairach_with_skull.lta and talairach.lta transforms (respectively). However, neither of those flags are options in the current public release of FS 4.0.5.
Doug: Will the forthcoming FS release have those flags available as options to tkregister2? We've found them to be a very helpful addition to tkregister2, and it sounds like Jeff would as well.
Jeff: In the interim, here are my notes on how to do what you want to accomplish (N.B. Doug's improved tkregister2 using the --gca-skull flag is a more powerful approach because it makes the target the subject space and thus the surfaces can also be meaningfully displayed via the --surf option):
mri_convert $FREESURFER_HOME/average/RB_all_withskull_2007-08-08.gca RB_gca.mgz --frame 0 (N.B. add --conform option if you want to avoid format conversion within tkregister2 itself (which slows loading))
tkregister2 --targ RB_gca.mgz --mov nu_noneck.mgz \ --lta transforms/talairach_with_skull.lta --reg tmp.reg
cheers, -Mike H.
On Mon, 2008-09-08 at 15:54 -0600, Jeff Sadino wrote:
Hello,
I am trying to validate the ICV value by lining up the nu_noneck image with the RB image. These are the commands I am using:
mri_convert nu_noneck.mgz --apply_transform transforms/talairach_with_skull.lta nu_noneck5.mgz tkmedit -f nu_noneck5.mgz & tkmedit -f /apps/freesurfer/average/RB_all_withskull_2006-02-15.gca &
When I compare these two, they don't line up. The RB image is all blurry (Maybe it's a combination of several subjects and is supposed to be like that). Any suggestions as to how to line up the noneck transformed imaged to the RB image is greatly appreciated!
I also attached three files to show what's happening in my transform process.
Thanks, Jeff
Date: Fri, 5 Sep 2008 20:31:55 -0400 From: fischl@nmr.mgh.harvard.edu To: jsadino@hotmail.com Subject: RE: [Freesurfer] Talairach_with_skull.lta accuracy
can you send this to the list so others can respond? Where did you get the tmp.reg file? If you have transformed them already you shouldn't need tkregister2 (or a registration file in general) to compare them.
On Fri, 5 Sep 2008, Jeff Sadino wrote:
Thank you very much for your help. Your knowledge and quick responses are very much appreciated.
I converted my image into transformed space, but it still does not match the RB image. These are the commands I used:
mri_convert nu_noneck.mgz --apply_transform
transforms/talairach_with_skull.lta nu_noneckt3.mgz
tkregister2 --targ
/apps/freesurfer/average/RB_all_withskull_2006-02-15.gca --reg tmp.reg --lta /data/040002_S04_bakj/mri/transforms/talairach_with_skull.lta --mov /data/040002_S04_bakj/mri/nu_noneckt3.mgz
Again, my main goal is to verify the ICV value, so to do that I need to verify that the lta file is transforming the nu_noneck image onto the RB image...I think.
What step am I missing?
Thank you Jeff
Date: Fri, 5 Sep 2008 13:09:21 -0400 From: fischl@nmr.mgh.harvard.edu To: jsadino@hotmail.com CC: greve@nmr.mgh.harvard.edu Subject: RE: [Freesurfer] Talairach_with_skull.lta accuracy
that won't generate a transformed volume, just display it in the transformed space. That is, the volumes you sent were in their native space and would have no reason to be aligned. The neck removal seems to have been fine in both cases. On Thu, 4 Sep 2008, Jeff Sadino wrote:
ps, sorry, I forgot to include my command:
tkregister2 --targ /apps/freesurfer/average/RB_all_withskull_2006-02-15.gca --mov /data/${file}/mri/nu_noneck.mgz --lta /data/${file}/mri/transforms/talairach_with_skull.lta --reg tmp.reg
I believe this will transform the nu_neck.mgz image via the talairach matrix, so that it matches the RB image.
btw, we processed the scans using freesurfer 3, but the tkregister2 that came with freesurfer 3 did not include the --lta tag, so I am using the tkregister2 from freesurfer4.
Thanks!
>Date: Thu, 4 Sep 2008 21:49:27 -0400 >From: fischl@nmr.mgh.harvard.edu >To: jsadino@hotmail.com >CC: freesurfer@nmr.mgh.harvard.edu >Subject: Re: [Freesurfer] Talairach_with_skull.lta accuracy > >Hi Jeff, > >why do you say that one is bad? We don't resample, so neither one will be >aligned with the atlas unless you apply the transform. > >We are aware that the eTIV measures have more variability than we would >like, and we're looking into it, but don't have a good solution from only >T1 images. > >cheers, >Bruce >On Thu, 4 Sep 2008, >Jeff Sadino wrote: > > > >>Hello all, >> >>According to aseg.stats, our ICV values fluctuate on the same subject across scans, sometimes by up to 10% I have been validating the talairach_with_skull.lta accuracy by comparing the nu_neck.mgz to the RB_all_withskull_2006-02-15.gca template. I read on the boards that it is not all that necessary that the brains match, that it is more important for the skulls to match. The RB is blurred (I don't fully understand where the RB comes from, I think from 22 subjects that the freesurfers combined into one?), so am I matching up the whitest areas of the RB to the skull on the nu_neck image? I noticed also that the bottom of the brains rarely match at all to the RB image...is this normal? Are there any other parts of the image that need to be checked for validity? >> >>I attached two nu_noneck.mgz files, one was good and the other bad. Are these images typical? >> >>Thank you for your help, >>Jeff >> >>_________________________________________________________________ >>See how Windows Mobile brings your life together?at home, work, or on the go. >>http://clk.atdmt.com/MRT/go/msnnkwxp1020093182mrt/direct/01/ >> >> _________________________________________________________________ Get more out of the Web. Learn 10 hidden secrets of Windows Live. http://windowslive.com/connect/post/jamiethomson.spaces.live.com-Blog-cns!55...
See how Windows connects the people, information, and fun that are part of your life. http://clk.atdmt.com/MRT/go/msnnkwxp1020093175mrt/direct/01/
See how Windows Mobile brings your life together—at home, work, or on the go. http://clk.atdmt.com/MRT/go/msnnkwxp1020093182mrt/direct/01/ _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
freesurfer@nmr.mgh.harvard.edu
-
Doug Greve -
Jeff Sadino -
Michael Harms