I'll leave the first question for Bruce or Jenni. The 2nd depends on what you are going to do with the talairach. If you are not going to be reporting talairach coords, then you can be quite sloppy as they are only used to create seed points for more-or-less finding the CC, pons, etc. If you are going to report talairach coords, then you only need to align the outside of the brain to the mni305. There is no perfect talairach reg, and in the best case it's going to be pretty awful:). If you get it within a centimeter, you're probably ok.

doug

Paul Greenberg wrote:

Hello,

I've got a question about cortical surfaces in the ventral regions below the basal ganglia and about tkregister output quality control.

I'm using the latest version of freesurfer: freesurfer-Linux-centos4_x86_64-stable-pub-v3.0.5

1. In the majority of my data sets, white matter surfaces near the

basal ganglia rise dorsally to include large volumes of the basal ganglia (putamen, nucleus accumbens, both hemispheres) - attached image. This results in inaccurate cortical volumes and thickness estimates from the surface based data for these regions. How can I compensate for these surface problems in order to get the best possible estimate of total cortical volume? I've tried editing the white matter manually and this has little effect on the locations of white matter surfaces in the basal ganglia. 2) I'm having a difficult time knowing how to determine if the talairach registration is good enough for each data set, and judging the quality of my manual edits using tkregister. Could you post an additional series of tiff or jpg images (multiple parasagittal, coronal, horizontal) showing how cortical surfaces and movable volumes are supposed to align in each plane? It would help to see examples of registrations that are "perfect", "good enough not to edit", and "bad enough to require adjustment". Thanks very much!,

Paul

---

---

Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer

Hi Paul,

are these regions properly labeled in the aseg? The surface is supposed to be fixed (i.e. not moving) in regions adjacent to subcortical structures. I guess the easiest thing to do is to create a label that only includes cortical regions and compute the volume limited to it (this is what the label/?h.cortex.label file is supposed to be - autogenerated by mris_make_surfaces).

cheers, Bruce

On Mon, 26 Mar 2007, Paul Greenberg wrote:

Hello,

I've got a question about cortical surfaces in the ventral regions below the basal ganglia and about tkregister output quality control.

I'm using the latest version of freesurfer: freesurfer-Linux-centos4_x86_64-stable-pub-v3.0.5

1. In the majority of my data sets, white matter surfaces near the basal

ganglia rise dorsally to include large volumes of the basal ganglia (putamen, nucleus accumbens, both hemispheres) - attached image. This results in inaccurate cortical volumes and thickness estimates from the surface based data for these regions. How can I compensate for these surface problems in order to get the best possible estimate of total cortical volume? I've tried editing the white matter manually and this has little effect on the locations of white matter surfaces in the basal ganglia. 2) I'm having a difficult time knowing how to determine if the talairach registration is good enough for each data set, and judging the quality of my manual edits using tkregister. Could you post an additional series of tiff or jpg images (multiple parasagittal, coronal, horizontal) showing how cortical surfaces and movable volumes are supposed to align in each plane? It would help to see examples of registrations that are "perfect", "good enough not to edit", and "bad enough to require adjustment". Thanks very much!,

Paul