Jenni Pacheco wrote:
Hi Martin,
Control points should fix the problem. Try adding some to this brighter region, as well as a few to regions currently labeled to 110. Sometimes this helps. After you've saved your control points (be sure to save them) then you should run:
recon-all -normalization -usecontrolpoints -subjid <subjid>
to make sure that it uses your control points. If you look at your brain T1 volume now, those bright spots should be taken care of. If so you'll need to re-run these steps: recon-all -skullstrip -subcortseg -normalization2 -segmentation -subjid <subjid>
If it hasn't worked, let me know, maybe there's another problem.
good luck, Jenni
On Fri, 28 Oct 2005, Martin Ystad wrote:
Hi, I'm currently running the latest developmental release for RedHat 9, but I seem to have some problems concerning the intesity-normalization. Unfortunately, my images have some intensity inhomogenities, but the first normalization-step seems to take care of most of these. However, some regions inside the white matter has a somewhat high value, for example 130 -140. These regions are not included in the white matter when I run the recon-all -segmentation program. Adding control-points to these areas does not work, obviously, since only values lower than 110 are considered when the normalization-step is re-runned. (right?) Is there any way to include these areas in my final white matter whithout manually drawing them onto the wm.mgz -volume? For instance raising the threshold for wm-segmentation? If so, would this produce any kind new artifacts which I need to be aware of?
Thanks,
Martin Ystad Medical Student University of Bergen Institute of Biomedicine Jonas Lies vei 91, 5009 Bergen, Norway.
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Hi, thanks for your help. I tried what you suggested and it turned out somewhat better. Still, there are large areas that are not included in the wm.mgz. The problem with some of the data sets is that the intensity inhomogenities in the center of the brain also reaches some very low values. I'm worried that if I include these with control-points, some of the gray matter will be misinterpreted as white matter. Is this a legitimate concern, or does the program work in a different way? If so, how low can the control point value be to avoid that this happens?
Ps. If I add a control point to one area with a specific high or low value, does this take care of all the other areas with similar values across the brain, or do I have to add a control point to all areas that have discrepancies in intensity?
Thanks, Martin
it only affects points that are close to it in space. Pretty much every point is corrected according to the bias field associated with the nearest control point, with some smoothing. Why are the values near the center of the brain so dark? Was this aquired with a surface coil?
cheers, Bruce
On Sun, 30 Oct 2005, Martin Ystad wrote:
Jenni Pacheco wrote:
Hi Martin,
Control points should fix the problem. Try adding some to this brighter region, as well as a few to regions currently labeled to 110. Sometimes this helps. After you've saved your control points (be sure to save them) then you should run:
recon-all -normalization -usecontrolpoints -subjid <subjid>
to make sure that it uses your control points. If you look at your brain T1 volume now, those bright spots should be taken care of. If so you'll need to re-run these steps: recon-all -skullstrip -subcortseg -normalization2 -segmentation -subjid
<subjid>
If it hasn't worked, let me know, maybe there's another problem.
good luck, Jenni
On Fri, 28 Oct 2005, Martin Ystad wrote:
Hi, I'm currently running the latest developmental release for RedHat 9, but I seem to have some problems concerning the intesity-normalization. Unfortunately, my images have some intensity inhomogenities, but the first normalization-step seems to take care of most of these. However, some regions inside the white matter has a somewhat high value, for example 130 -140. These regions are not included in the white matter when I run the recon-all -segmentation program. Adding control-points to these areas does not work, obviously, since only values lower than 110 are considered when the normalization-step is re-runned. (right?) Is there any way to include these areas in my final white matter whithout manually drawing them onto the wm.mgz -volume? For instance raising the threshold for wm-segmentation? If so, would this produce any kind new artifacts which I need to be aware of?
Thanks,
Martin Ystad Medical Student University of Bergen Institute of Biomedicine Jonas Lies vei 91, 5009 Bergen, Norway.
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Hi, thanks for your help. I tried what you suggested and it turned out somewhat better. Still, there are large areas that are not included in the wm.mgz. The problem with some of the data sets is that the intensity inhomogenities in the center of the brain also reaches some very low values. I'm worried that if I include these with control-points, some of the gray matter will be misinterpreted as white matter. Is this a legitimate concern, or does the program work in a different way? If so, how low can the control point value be to avoid that this happens?
Ps. If I add a control point to one area with a specific high or low value, does this take care of all the other areas with similar values across the brain, or do I have to add a control point to all areas that have discrepancies in intensity?
Thanks, Martin _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Bruce Fischl wrote:
it only affects points that are close to it in space. Pretty much every point is corrected according to the bias field associated with the nearest control point, with some smoothing. Why are the values near the center of the brain so dark? Was this aquired with a surface coil?
cheers, Bruce
On Sun, 30 Oct 2005, Martin Ystad wrote:
Jenni Pacheco wrote:
Hi Martin,
Control points should fix the problem. Try adding some to this brighter region, as well as a few to regions currently labeled to 110. Sometimes this helps. After you've saved your control points (be sure to save them) then you should run:
recon-all -normalization -usecontrolpoints -subjid <subjid>
to make sure that it uses your control points. If you look at your brain T1 volume now, those bright spots should be taken care of. If so you'll need to re-run these steps: recon-all -skullstrip -subcortseg -normalization2 -segmentation -subjid <subjid>
If it hasn't worked, let me know, maybe there's another problem.
good luck, Jenni
On Fri, 28 Oct 2005, Martin Ystad wrote:
Hi, I'm currently running the latest developmental release for RedHat 9, but I seem to have some problems concerning the intesity-normalization. Unfortunately, my images have some intensity inhomogenities, but the first normalization-step seems to take care of most of these. However, some regions inside the white matter has a somewhat high value, for example 130 -140. These regions are not included in the white matter when I run the recon-all -segmentation program. Adding control-points to these areas does not work, obviously, since only values lower than 110 are considered when the normalization-step is re-runned. (right?) Is there any way to include these areas in my final white matter whithout manually drawing them onto the wm.mgz -volume? For instance raising the threshold for wm-segmentation? If so, would this produce any kind new artifacts which I need to be aware of?
Thanks,
Martin Ystad Medical Student University of Bergen Institute of Biomedicine Jonas Lies vei 91, 5009 Bergen, Norway.
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Hi, thanks for your help. I tried what you suggested and it turned out somewhat better. Still, there are large areas that are not included in the wm.mgz. The problem with some of the data sets is that the intensity inhomogenities in the center of the brain also reaches some very low values. I'm worried that if I include these with control-points, some of the gray matter will be misinterpreted as white matter. Is this a legitimate concern, or does the program work in a different way? If so, how low can the control point value be to avoid that this happens?
Ps. If I add a control point to one area with a specific high or low value, does this take care of all the other areas with similar values across the brain, or do I have to add a control point to all areas that have discrepancies in intensity?
Thanks, Martin _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Hi, Yes, we used an 8-channel surface coil from GE. I've attached a picture from one of the scans to illustrate.
This pattern of light-dark seems to follow a sinusoidal-curve across the entire volume in the saggital direction. Is there another way of fixing this in freesurfer without using control-points?
Thanks, Martin
Hi,
mris_gm does the group compariosn of cortical thickness vertex by vertex (to my understanding). I have a question, What is the size of vertex during the test? is it voxel by voxel?
Thanks,
Antao
Hi Antao,
the vertices are distributed at approximately even 1mm intervals over the surface.
cheers, Bruce
On Mon, 31 Oct 2005, Antao Du wrote:
Hi,
mris_gm does the group compariosn of cortical thickness vertex by vertex (to my understanding). I have a question, What is the size of vertex during the test? is it voxel by voxel?
Thanks,
Antao
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
freesurfer@nmr.mgh.harvard.edu
-
Antao Du -
Bruce Fischl -
Martin Ystad