External Email - Use Caution
Hi Jack,
it depends on how different your questions are and how many cases you have in the cross study vs long study.
If the questions are different or you have a large cohort for the cross sectional, then it makes sense to keep the two analyses separately. You would then include all cases that were processed with simple recon-all into the cross study, and all cases with multiple time points - processed with the longitudinal pipeline - in the longitudinal analysis.
If questions are related and there is a lot of overlap (many cases with multiple time points), then you can also include all data into a single LME model, also the single-time-point cases. LME can use this information to estimate cross subject variances better. For this it is important to run the single-time-point cases through the longitudinal pipeline! This is important so that they undergo the same processing steps as all the other data (after cross recon-all , create a base with only one time point and then run the long step with that time point and base-template).
Best, Martin
On 14. Jul 2023, at 00:56, Jackson Lee jacksonml@student.unimelb.edu.au wrote:
Hi Yujing,
Thanks for your reply – I have a follow up query (see email thread below):
I have both a cross-sectional aim and a longitudinal aim I am hoping to explore with FreeSurfer. While they are separate aims, some participants are present in both the cross-sectional and longitudinal arm. Currently I have run recon-all on all participants for my study using the standard recon-all command. As mentioned, I am hoping to do hippocampal subfield analysis (both cross-sectional and longitudinal).
I have read both the hippocampal subfield and longitudinal pipelines available from FreeSurfer (linked below):
I was planning on running the longitudinal recon-all pipeline as it seems that running longitudinal pipeline allows for greater reliability as "the subject is used as their own control". But if I do this, then the subjects that are also in the cross-sectional arm will potentially have different outputs (even though they are the same scan) compared to outputs from the longitudinal pipeline. This would be the same for the subfield analysis.
What is the best approach here? Would I be better off running all participants (even those with one T1-weighted scan) with the longitudinal pipeline so that every participant is treated the homogenously?
Let me know if I need to clarify anything!
Warm regards, Jack
Hi Jack,
Longitudinal recon-all pipeline is available in Freesurfer 7.3.2.
To do longitudinal segmentation using the hippocampal subfield pipeline, it is a requirement that the data have been processed with the longitudinal stream Freesurfer.
Best,
Yujing
From: freesurfer-bounces@nmr.mgh.harvard.edumailto:freesurfer-bounces@nmr.mgh.harvard.edu <freesurfer-bounces@nmr.mgh.harvard.edumailto:freesurfer-bounces@nmr.mgh.harvard.edu> On Behalf Of Jackson Lee Sent: Wednesday, July 12, 2023 10:25 PM To: freesurfer@nmr.mgh.harvard.edumailto:freesurfer@nmr.mgh.harvard.edu Subject: [Freesurfer] Longitudinal pipeline for FreeSurfer (v7.3.2)
External Email - Use Caution
Hi there,
I am hoping to process some longitudinal data using the longitudinal recon-all pipeline available in FreeSurfer: MailScanner has detected a possible fraud attempt from "secure-web.cisco.com" claiming to behttps://surfer.nmr.mgh.harvard.edu/fswiki/LongitudinalChangeLoghttps://secure-web.cisco.com/1Y9uqixBqGC2B0XHSMg2zr4QS96oUVSkkZcMdhRqdEkVD1KblmslkRndxPSvHW3LBNES_IqNsfYFPgJ0qwhvc1bNFhpnjhfJbHFnw_JsG65hDnTYTh0tH7f-TFks6LnM0ShLPbcTaqRM3Ot0icc5Qz_q6LdCD7ArrMeBy-ZAdoU36C3aO2YWeJUSYV7V1ziGk9n8wLoM3ExyKYNOgnUc1ECoCPAxei2kkH5NycZEWbOU3lH1BrNSjGuSyBOv-39-ZdPFU28SYlq9ASUOR3nT6_1bci2ZeViPVJ1yWa4fuw2q150BmfUOtkpMHpCeEh8c_XtRYiQTDNa9dwtnGNQeSNA/https%3A%2F%2Fsurfer.nmr.mgh.harvard.edu%2Ffswiki%2FLongitudinalChangeLog
Will this pipeline work for FreeSurfer v7.3.2? I cannot see any resources to suggest this is the case.
Additionally, I am hoping to conduct some longitudinal analysis using the hippocampal subfield pipeline: MailScanner has detected a possible fraud attempt from "secure-web.cisco.com" claiming to behttps://surfer.nmr.mgh.harvard.edu/fswiki/HippocampalSubfieldsAndNucleiOfAmy...https://secure-web.cisco.com/1ILqdmv74otr5LeRS7UQz83p5UDddkh5pv-Luvs5PHnW-6BfivGurDffrJ3MLUO6_g51fdQTLLVgIlCbkxz5HvAXXSVsYl5DXzxX7vuZYuXle0lC8SiAlMxablGmyrpj5kLiT4ur9kz0mN3fPEMReUKCskogc5WXK6BMHzhCAT8xCcztE2TUmLJDn6PZN5XiYr_4CE3ZPhwUrzW3qtPXA6aBDo11WlFO7kBKH5exeJUJ2i66Ot29HyeIuOKUQ-suEyIZKsZQ4iyRJMNCLSXJpr-ryBfctpn-E6mEynv9KcCJmJG3V6eJtPJnv5EdktzlUirApmR8nLA-iMVNLDVlJAg/https%3A%2F%2Fsurfer.nmr.mgh.harvard.edu%2Ffswiki%2FHippocampalSubfieldsAndNucleiOfAmygdala Would the best course of action be to run recon-all with the longitudinal pipeline first (assuming it is applicable for freeSurfer v7.3.2) and then use the longitudinal command for post-hoc subfield analysis?: segmentHA_T1_long.sh <baseID> [SUBJECTS_DIR]
Thank you, Jack
_______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://secure-web.cisco.com/1V6ziip-kZFPCA4cs7RsqRVs6xkHn9GSonoXcQ4TTfgsPWE...