Dear Dr Bruce Fischl.
How are you? I preprocessed 41 subjects' MRI to measure the volumes of the cortical and subcortical structures with ver 3.04 Freesurfer. What kind of atlas (Desikan-Killiany atlas or Destrieux atlas?) was used for my data? And how can I choose the atlas between the two atlases.
If you provide me the tool for new training set, I'd like to create my own study specific atlas. Really thank for your reply!
Sincerely
Woo Suk, Tae
-----Original Message----- From: Bruce Fischl [mailto:fischl@nmr.mgh.harvard.edu] Sent: Monday, October 01, 2007 9:03 PM To: Woo Suk, Tae Cc: freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] automated parcellation: training set of Freesurfer
there are two training sets we use for the two difference parcellation schemes. Once consists of about 40 subjects distributed into 4 groups (young, middle aged, elderly and Alzheimer's for the Desikan-Killiany atlas). The other is 14 young and middle aged subjects in the Destrieux atlas. Defining a new training set is extremely labor intensive, but we distrubute tools for you to do so if you decide to. cheers, Bruce On Mon, 1 Oct 2007, Woo Suk, Tae wrote:
Dear experts.
In the previous technical papers of Freesurfer, the the method of automated parcellation was described,
"This procedure automatically assigns a neuroanatomical label to each voxel in an MRI volume based on probabilistic information automatically estimated from manually labeled training set."
I'd like to know more detail, how many training set was used in Freesurfer, and is it possible to define the study specific training set for automated parcellation using Freesurfer?
Sincerely
Woo Suk, Tae Seoul, Korea
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
by default we run both parcellations, so you should have them. Which to choose should be based on what your questions are. The Desikan-Killiany has in general larger units than the Destrieux one, but also anatomically different.
As for your own training set, you'll need to create a set of manual labels and save them in .annot files, then run mris_ca_train on them. It should be documented on our wiki.
cheers, Bruce
On Tue, 2 Oct 2007, [EUC-KR] "Woo Suk, Tae" wrote:
Dear Dr Bruce Fischl.
How are you? I preprocessed 41 subjects' MRI to measure the volumes of the cortical and subcortical structures with ver 3.04 Freesurfer. What kind of atlas (Desikan-Killiany atlas or Destrieux atlas?) was used for my data? And how can I choose the atlas between the two atlases.
If you provide me the tool for new training set, I'd like to create my own study specific atlas. Really thank for your reply!
Sincerely
Woo Suk, Tae
-----Original Message----- From: Bruce Fischl [mailto:fischl@nmr.mgh.harvard.edu] Sent: Monday, October 01, 2007 9:03 PM To: Woo Suk, Tae Cc: freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] automated parcellation: training set of Freesurfer
there are two training sets we use for the two difference parcellation schemes. Once consists of about 40 subjects distributed into 4 groups (young, middle aged, elderly and Alzheimer's for the Desikan-Killiany atlas). The other is 14 young and middle aged subjects in the Destrieux atlas. Defining a new training set is extremely labor intensive, but we distrubute tools for you to do so if you decide to. cheers, Bruce On Mon, 1 Oct 2007, Woo Suk, Tae wrote:
Dear experts.
In the previous technical papers of Freesurfer, the the method of automated parcellation was described,
"This procedure automatically assigns a neuroanatomical label to each voxel in an MRI volume based on probabilistic information automatically estimated from manually labeled training set."
I'd like to know more detail, how many training set was used in Freesurfer, and is it possible to define the study specific training set for automated parcellation using Freesurfer?
Sincerely
Woo Suk, Tae Seoul, Korea
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
freesurfer@nmr.mgh.harvard.edu
-
"Woo Suk, Tae" -
Bruce Fischl