External Email - Use Caution
Hi experts, We want to combine two large datasets (of patients with different diseases) that have been processed and manually edited using Freesurfer 5.3 and Freesurfer 6.0. From the Freesurfer website and previous posts, we have read that we should re-run the Freesurfer 5.3 data that has been manually edited (both brain masks and pial surfaces) through Freesurfer 6.0 using recon-all and using the -continued flag, and that the manual edits from 5.3 should carry over. However, we have found that around one-third of subjects have skull stripping errors after rerunning through 6.0. Has anyone else found this? Have we missed something? A related question: We have heard it argued that if the data has been manually edited and quality checked, the version of Freesurfer which it is run through should not matter, therefore it is ok to combine data from different versions – what do you think of this argument? An additional argument along those lines that we have heard is: If one were to re-process data with a later version of Freesurfer, using edits from an older version would introduce biases from that version – what are your thoughts on this argument? Thank you.
Lisanne
On 2/18/2021 9:52 AM, Lisanne Michelle Jenkins wrote:
External Email - Use Caution
Hi experts,
We want to combine two large datasets (of patients with different diseases) that have been processed and manually edited using Freesurfer 5.3 and Freesurfer 6.0.
From the Freesurfer website and previous posts, we have read that we should re-run the Freesurfer 5.3 data that has been manually edited (both brain masks and pial surfaces) through Freesurfer 6.0 using recon-all and using the -continued flag, and that the manual edits from 5.3 should carry over.
However, we have found that around one-third of subjects have skull stripping errors after rerunning through 6.0. Has anyone else found this? Have we missed something?
I have not noticed this. Are you saying that your edits are not carrying over or that there are new errors in 6.0 that were not in 5.3?
A related question: We have heard it argued that if the data has been manually edited and quality checked, the version of Freesurfer which it is run through should not matter, therefore it is ok to combine data from different versions – what do you think of this argument?
This is not a question of manual editing. Different versions can have different parameters/algorithms that can create subtle but systematic differences in thickness. If all of one group came from one version and all of another group came from a different version, then I think you could see differences due to version
An additional argument along those lines that we have heard is: If one were to re-process data with a later version of Freesurfer, using edits from an older version would introduce biases from that version – what are your thoughts on this argument?
I doubt it.
Thank you.
Lisanne
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
External Email - Use Caution
Hi, yes in some cases there were new skull stripping errors in 6.0 that were not present in 5.3. In other cases where images were perhaps under-edited after 5.3 processing, these errors were more pronounced after 6.0 processing.
And yes, thank you for confirming what I had suspected!
Thanks, Lisanne
From: freesurfer-bounces@nmr.mgh.harvard.edu on behalf of "Douglas N. Greve" dgreve@mgh.harvard.edu Reply-To: Freesurfer support list freesurfer@nmr.mgh.harvard.edu Date: Monday, February 22, 2021 at 9:18 AM To: "freesurfer@nmr.mgh.harvard.edu" freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] Questions related to combining datasets run through different versions
On 2/18/2021 9:52 AM, Lisanne Michelle Jenkins wrote:
External Email - Use Caution Hi experts, We want to combine two large datasets (of patients with different diseases) that have been processed and manually edited using Freesurfer 5.3 and Freesurfer 6.0. From the Freesurfer website and previous posts, we have read that we should re-run the Freesurfer 5.3 data that has been manually edited (both brain masks and pial surfaces) through Freesurfer 6.0 using recon-all and using the -continued flag, and that the manual edits from 5.3 should carry over. However, we have found that around one-third of subjects have skull stripping errors after rerunning through 6.0. Has anyone else found this? Have we missed something? I have not noticed this. Are you saying that your edits are not carrying over or that there are new errors in 6.0 that were not in 5.3?
A related question: We have heard it argued that if the data has been manually edited and quality checked, the version of Freesurfer which it is run through should not matter, therefore it is ok to combine data from different versions – what do you think of this argument? This is not a question of manual editing. Different versions can have different parameters/algorithms that can create subtle but systematic differences in thickness. If all of one group came from one version and all of another group came from a different version, then I think you could see differences due to version
An additional argument along those lines that we have heard is: If one were to re-process data with a later version of Freesurfer, using edits from an older version would introduce biases from that version – what are your thoughts on this argument? I doubt it.
Thank you.
Lisanne
_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edumailto:Freesurfer@nmr.mgh.harvard.edu
https://secure-web.cisco.com/1d3TQ2b9G7N7mMJfwHDD0M0jwxzQ_kQKVJxZ6wT9KcL2Whj...https://secure-web.cisco.com/1fqU3OzvSXQkNVVtFt4T_94id7JkWAEdLjApRMyTroQxzviB2BuJVPrijsgKAT0fMPnIcFNgnSouIzH4mBCHyhJOcT0yNw0M2HZBJ_IakaXldiUihrwlfZfELWwcIirUCwh_pd0ZGk-JqmMDpEA9N6yYokBASd7_DP5HzXk337fu2PL7EMzHf3AT4rt4Rr-njjQxuFxSNt0uA3WxnuIqaUOrTiVvbd2WLa0AF-O6e-_pr-0qA2uxwmVfPeH_AIrhFQSK75UGumV5SCkBhK6oXmg/https%3A%2F%2Furldefense.com%2Fv3%2F__https%3A%2Fmail.nmr.mgh.harvard.edu%2Fmailman%2Flistinfo%2Ffreesurfer__%3B%21%21Dq0X2DkFhyF93HkjWTBQKhk%21FziCoqioNjskVIA5x0RhD31Xg6-UVUaQiABscN-CIoUVQk4finmrrZqEUtjIkc1PhF9cRpP7Kjw%24
freesurfer@nmr.mgh.harvard.edu
-
Douglas N. Greve -
Lisanne Michelle Jenkins