External Email - Use Caution

My opinion (the reviewers of your manuscript may have a different one): In general it is fine to use scans which are 10 years old. But in your case, the problem is that all control subjects were scanned in batch 1 (date: 10y ago/scanner 1/protocol 1/t2-availabilty=no/...) and all case subjects with batch 2 (date: now/scanner 2/protocol 2/t2-availabilty=yes/...).

If half of your scans would be 10 years old (including a random mixture of cases and controls) and the other half new with t2 (also including a random mixture of cases and controls), I guess I would not care *that* much. But this way, there is no way for the statistical analysis to differentiate between batch effect and group effect.

Therefore, I doubt that you would be able to publish whatever results you get in a decent journal.

Tim

-- Dr. Tim Schäfer Postdoc Computational Neuroimaging Department of Child and Adolescent Psychiatry, Psychosomatics and Psychotherapy University Hospital Frankfurt, Goethe University Frankfurt am Main, Germany

On 02/15/2022 6:11 AM Eszter Boros boros.eszter0226@gmail.com wrote:

External Email - Use Caution

Thank you. This is very helpful. And just a follow-up question about this: Can we use data that old? (And make a comparison to images collected this year). I mean is it possible to compare structural scans which were acquired such a long time ago and with a different scanner using Freesurfer? (e.g., measuring cortical thickness)

All the best, Eszter Boros

Douglas N. Greve dgreve@mgh.harvard.edu ezt írta (időpont: 2022. febr. 14., H, 10:48):

...

You will definitely get systematic differences between using T2 and not, so you could have a false positive (or negative) if you are comparing those two groups. On the other hand, you probably changed the T1 protocol a lot between now and 10 years ago, so you are likely to have bias either way.

On 2/11/2022 12:53 PM, Eszter Boros wrote:

    External Email - Use Caution

Dear FreeSurfer Team,

We have collected T1 and T2 images in our study and running recon-all with T2 gives us much better results (it was really good to see it).

Our aim is to compare these structural data (in Group A) to another patient group (Group B). The problem is that the data of Group B was collected 10 years ago and we did not have a T2 in that study.

I am wondering what would be the best approach? Maybe we should not use T2 in Group A because we do not have a T2 in Group B? Or if we do the manual corrections correctly in Group B, this should not be a problem? I am using Freesufer 7.1.1

Thank you for your answer! I greatly appreciate it.

All the best, Eszter Boros

---

Freesurfer mailing listFreesurfer@nmr.mgh.harvard.eduhttps://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer

---

Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://secure-web.cisco.com/1wVcWXiNn37WyoR-pT_kwIeNB_UwOvN3uE9E1Dd3XP-QqzG...

---

Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://secure-web.cisco.com/1E1Bw_s0DBBQ9GQu4ddAQDpfCiHBC_PImT8rnYQqf9rEV2Y...