External Email - Use Caution
Our group is trying to decide on whether we should move to FS8/clinical recon all indefinitely, however we note some inconsistencies. In some instances, SynthSeg seems to completely ignore the underlying structure and “imputes” tissue. Cortical thickness estimates are overall most different from FS7, as may be expected, however, in some cases volumes (e.g., hippocampus) may also be overestimated.
We have 1mm isotropic T1s, which I see that recon-all may be "generally more accurate than recon-all-clinical.” We’re looking to take advantage of SynthSeg’s ability to overcome poor data quality/dropout/complex folding etc., but do not want to necessarily “in-paint normal looking tissue.” Can you provide any additional information or guidance on the best path forward?
Thank you very much for your time and any guidance!
________________________________
This electronic message is intended only for the use of the individual(s) and entity named as recipients in the message. If you are not an intended recipient of this message, please notify the sender immediately and delete the material from any computer. Do not deliver, distribute or copy this message, and do not disclose its contents or take any action in reliance on the information it contains. Thank you.
Dear Marc,
* Cortically: If you have 1mm isotropic T1s, then recon-all is generally more accurate than recon-all-clinical, since its designed explicitly for such images (whereas recon-all-clinical is resolution- and modality-agnostic). BUT: if you have big lesions (as is some of the examples you sent), then recon-all-clinical may be more robust, as it will try to “inpaint” normal looking tissue. Of course, if that’s not what you want, then revert to recon-all if you can 😉 * Subcortically: FS8 uses SynthSeg so it shouldn’t matter much.
Cheers, /Eugenio
--
Juan Eugenio Iglesias
From: Rudolph, Marc D Marc.Rudolph@advocatehealth.org Date: Friday, November 7, 2025 at 10:47 AM To: freesurfer@nmr.mgh.harvard.edu freesurfer@nmr.mgh.harvard.edu Subject: [Freesurfer] Recon-all-clinical/SynthSeg Concerns w/ Atrophy
External Email - Use Caution Our group is trying to decide on whether we should move to FS8/clinical recon all indefinitely, however we note some inconsistencies. In some instances, SynthSeg seems to completely ignore the underlying structure and “imputes” tissue. Cortical thickness estimates are overall most different from FS7, as may be expected, however, in some cases volumes (e.g., hippocampus) may also be overestimated.
We have 1mm isotropic T1s, which I see that recon-all may be "generally more accurate than recon-all-clinical.” We’re looking to take advantage of SynthSeg’s ability to overcome poor data quality/dropout/complex folding etc., but do not want to necessarily “in-paint normal looking tissue.” Can you provide any additional information or guidance on the best path forward?
Thank you very much for your time and any guidance! ________________________________
This electronic message is intended only for the use of the individual(s) and entity named as recipients in the message. If you are not an intended recipient of this message, please notify the sender immediately and delete the material from any computer. Do not deliver, distribute or copy this message, and do not disclose its contents or take any action in reliance on the information it contains. Thank you.
External Email - Use Caution
Thank you for your reply. In some cases, these are not large lesions. Unless I’m misunderstanding, we wouldn’t want to generate and/or trust surface/volume metrics. From your responses, it may be best for now to stick with recon-all. Again, thank you for your replies, Eugenio. Best, Marc
From: Iglesias Gonzalez, Juan E. JIGLESIASGONZALEZ@mgh.harvard.edu Date: Friday, November 7, 2025 at 10:54 AM To: Freesurfer support list freesurfer@nmr.mgh.harvard.edu Subject: [EXTERNAL] [Freesurfer] Re: Recon-all-clinical/SynthSeg Concerns w/ Atrophy
WARNING: This email originated from outside of Advocate Health (freesurfer-bounces@nmr.mgh.harvard.edu). ALWAYS use caution with links and attachments even if you trust the sender. NEVER provide your login information to anyone. USE Squish the Phish to report suspicious email.
Dear Marc,
* Cortically: If you have 1mm isotropic T1s, then recon-all is generally more accurate than recon-all-clinical, since its designed explicitly for such images (whereas recon-all-clinical is resolution- and modality-agnostic). BUT: if you have big lesions (as is some of the examples you sent), then recon-all-clinical may be more robust, as it will try to “inpaint” normal looking tissue. Of course, if that’s not what you want, then revert to recon-all if you can 😉 * Subcortically: FS8 uses SynthSeg so it shouldn’t matter much.
Cheers, /Eugenio
--
Juan Eugenio Iglesias
http://secure-web.cisco.com/15oeRRUC9BnPmZI8ndCxSWDJ2dr4-TBG44EZ5EhwvGMwzM93...https://secure-web.cisco.com/1X9vn0lkck79rgQR12QDnKCa5RUiR5uH05ISVnv3BaX-9jOjwM89MjCxlBAfaE0PyL6QKwNk9hdrTFy-1WHLgjyrLgouXnFgSyU0MhbVYzptefrx4JXh-REEHYJ07PXCUIZfRS4QGWEuxM-T98HokL8d8x6FZUu4ixTQ2RjWvM6E7dZ0YVs4aWeCd0Ljm0USVVuiS_SyfEhSb2redg91g7TM1hhMUM1sLb2RUN311WPkkUpVkgmVOZglBMN3eGH5-orqOyQXkT043nfX_D5iWR2Nc2mXRrywiQeA3h0pkWTRCJCMhla2qmxU_EWbpMTG9G7CTteONXxEYh24-XSEuGA/https%3A%2F%2Furldefense.com%2Fv3%2F__http%3A%2F%2Fwww.jeiglesias.com__;!!Kv7QgGdTlhIaqSqT!MCh4fP8Ph1ldFoGPe-t8E8sfJ87yv2Hc0uWbU2V1k-QXvEYAZp95Wt2NagwLUwFfQgNo9tpqDZBrPzSIUHDMFnJT0WJX0rtkgAJ07eil$
From: Rudolph, Marc D Marc.Rudolph@advocatehealth.org Date: Friday, November 7, 2025 at 10:47 AM To: freesurfer@nmr.mgh.harvard.edu freesurfer@nmr.mgh.harvard.edu Subject: [Freesurfer] Recon-all-clinical/SynthSeg Concerns w/ Atrophy
External Email - Use Caution Our group is trying to decide on whether we should move to FS8/clinical recon all indefinitely, however we note some inconsistencies. In some instances, SynthSeg seems to completely ignore the underlying structure and “imputes” tissue. Cortical thickness estimates are overall most different from FS7, as may be expected, however, in some cases volumes (e.g., hippocampus) may also be overestimated.
We have 1mm isotropic T1s, which I see that recon-all may be "generally more accurate than recon-all-clinical.” We’re looking to take advantage of SynthSeg’s ability to overcome poor data quality/dropout/complex folding etc., but do not want to necessarily “in-paint normal looking tissue.” Can you provide any additional information or guidance on the best path forward?
Thank you very much for your time and any guidance! ________________________________
This electronic message is intended only for the use of the individual(s) and entity named as recipients in the message. If you are not an intended recipient of this message, please notify the sender immediately and delete the material from any computer. Do not deliver, distribute or copy this message, and do not disclose its contents or take any action in reliance on the information it contains. Thank you.
________________________________
This electronic message is intended only for the use of the individual(s) and entity named as recipients in the message. If you are not an intended recipient of this message, please notify the sender immediately and delete the material from any computer. Do not deliver, distribute or copy this message, and do not disclose its contents or take any action in reliance on the information it contains. Thank you.
freesurfer@nmr.mgh.harvard.edu
-
Iglesias Gonzalez, Juan E. -
Rudolph, Marc D