If you have one subject, then there is no need to run a two stage model. Instead, create a design matrix or FSGD for that one subject. Eg, create the FSGD with one class. Create one Variable called TimePoint. Create an Input line for each time point indicating the class and the time point number as the variable (remove the mean of the time points from the time point number number). Then create a contrast with 0 1. When you run mri_glmfit, it will test the slope of the thickness-time effect (I'm assuming that is what you want).

On 4/17/2025 5:37 PM, Isabella Rossellini wrote:

     External Email - Use Caution

Dear Douglas,

I am sorry for the late response. I did not see your message up until now. I would like to see how the brain changes for a patient with anxiety who got better gradually (again, we have 5-time points (0, 3, 6, 9, 12). And then do the same for another participant who did not get better up until 12 months. That is the reason I would like to look at patients one by one.

I ran the mri_glmfit analysis but I received an error message:

ERROR: DOF=0

Is this because I tried to run it on one participant?

Sincerely, Isa Rosselini

Date: Mon, 3 Feb 2025 09:20:40 -0500 From: "Douglas N. Greve" dgreve@mgh.harvard.edu Subject: Re: [Freesurfer] longitudinal analysis on one subject To: freesurfer@nmr.mgh.harvard.edu Message-ID: ebc3545a-9335-4b43-b7ae-da3c0b1d0bdb@mgh.harvard.edu Content-Type: text/plain; charset="utf-8"

The glm can be run on this data in theory. With only 5 time points, you probably won't have much power though. What are you trying to test?

On Sat, Feb 1, 2025 at 11:56 PM Isabella Rossellini irossellini86@gmail.com wrote:

...

Hello Freesurfer users/experts,

I am working with Freesurfer 7.4.1 on a MacStudio (macOS 15.2) and I would like to ask your opinion about the following.

Basically, we had an anxiety treatment study and I would like to analyze one subject who has 5-time points (0, 3, 6, 9, 12) with a longitudinal pipeline, two-stage model. Is it possible to run this on one person only and see the changes between these time points?

I have never done this before on one subject so I am not sure what the best approach would be to look at the changes. After running the longitudinal preprocessing (base, long), should I generate the tables of freesurfer parcellation/segmentation stats data (aparcstats2table, asegstats2table)? Or can I just follow the two stage model steps? Maybe glm only works on a group of subjects. https://secure-web.cisco.com/1RCkpkv-7j7POY3iFHgRmXEUWgfCzKlqbBUGWLnc2RlZdY5...

(We have 9 more subjects - 10 in total - so I was hoping to visualize the pattern of changes one by one, in each subject, since these patients showed different changes in anxiety based on self-reported questionnaires)

Sincerely, Isa Rosselini

---

Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer

On 6/20/2025 3:19 PM, Isabella Rossellini wrote:

     External Email - Use Caution

Dear Douglas,

Thank you for your kind help and advice! Good to know that a two-stage model is not needed for one subject. So I should not run the longitudinal preprocessing either, right?

Actually, you can run it on a single subject as this will impose similar processing (eg, interpolation) that the longitudinal stream would have imposed.

Following your advice, I created this FSGD file below for subjid1 with 5 timepoints (0, 3, 6, 9, 12). Name of the imaging files: subjid1_base, subjid1_3mo, subjid1_6mo, subjid1_9mo, subjid1_12mo

GroupDescriptorFile 1 Title Onesubject Class Class1 Variables TimePoint Input subjid1_base Class1 0 Input subjid1_3mo Class1 3 #Input subjid1_6mo Class1 6 Input subjid1_9mo Class1 9 Input subjid1_12mo Class1 12

Does this look okay to you?

You also wrote that I should "remove the mean of the time points from the time point number number". (The mean is 6 in our case) I wonder why? Does this mean that the TimePoint variables should be -6 -3 0 3 6 instead? (and not 0 3 6 9 12)

Yes, this is correct. Removing the mean can make the design matrix better conditioned. It will also influence the offset; but if you're only interested in the slope, that part won't make a difference.

GroupDescriptorFile 1 Title Onesubject Class Class1 Variables TimePoint Input subjid1_base Class1 -6 Input subjid1_3mo Class1 -3 #Input subjid1_6mo Class1 0 Input subjid1_9mo Class1 3 Input subjid1_12mo Class1 6

Sincerely, Isa Rosselini

On Thu, Apr 17, 2025 at 5:37 PM Isabella Rossellini irossellini86@gmail.com wrote:

...

Dear Douglas,

I am sorry for the late response. I did not see your message up until now. I would like to see how the brain changes for a patient with anxiety who got better gradually (again, we have 5-time points (0, 3, 6, 9, 12). And then do the same for another participant who did not get better up until 12 months. That is the reason I would like to look at patients one by one.

I ran the mri_glmfit analysis but I received an error message:

ERROR: DOF=0

Is this because I tried to run it on one participant?

Sincerely, Isa Rosselini

Date: Mon, 3 Feb 2025 09:20:40 -0500 From: "Douglas N. Greve" dgreve@mgh.harvard.edu Subject: Re: [Freesurfer] longitudinal analysis on one subject To: freesurfer@nmr.mgh.harvard.edu Message-ID: ebc3545a-9335-4b43-b7ae-da3c0b1d0bdb@mgh.harvard.edu Content-Type: text/plain; charset="utf-8"

The glm can be run on this data in theory. With only 5 time points, you probably won't have much power though. What are you trying to test?

On Sat, Feb 1, 2025 at 11:56 PM Isabella Rossellini irossellini86@gmail.com wrote:

...

Hello Freesurfer users/experts,

I am working with Freesurfer 7.4.1 on a MacStudio (macOS 15.2) and I would like to ask your opinion about the following.

Basically, we had an anxiety treatment study and I would like to analyze one subject who has 5-time points (0, 3, 6, 9, 12) with a longitudinal pipeline, two-stage model. Is it possible to run this on one person only and see the changes between these time points?

I have never done this before on one subject so I am not sure what the best approach would be to look at the changes. After running the longitudinal preprocessing (base, long), should I generate the tables of freesurfer parcellation/segmentation stats data (aparcstats2table, asegstats2table)? Or can I just follow the two stage model steps? Maybe glm only works on a group of subjects. https://secure-web.cisco.com/1PBrCxPtnV_cyA4wjTbApPKLTn7-VE-boLvlA7ZOq8U2_gM...

(We have 9 more subjects - 10 in total - so I was hoping to visualize the pattern of changes one by one, in each subject, since these patients showed different changes in anxiety based on self-reported questionnaires)

Sincerely, Isa Rosselini

---

Freesurfer mailing list -- freesurfer@nmr.mgh.harvard.edu To unsubscribe send an email to freesurfer-leave@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman3/lists/freesurfer@nmr.mgh.harvard.e...

On 7/10/2025 3:51 PM, Isabella Rossellini wrote:

     External Email - Use Caution

Thank you! I found some significant results/regions in the negative direction. Does this mean that time impacts cortical thickness? So, there is a time x thickness effect, and with longer time/treatment, the cortical thickness decreases in those regions?

Yes, though it is usually not stated as "time-by-thickness effect" since time is an independent variable and thickness is a dependent variable.

Additionally, I wanted to visualize the thickness changes over time. In other words, I was hoping to extract and visualize cortical thickness values at these time points. What would be the best way to do this?

You could make an average at each time point.

Sincerely, Isa Rosselini

On Fri, Jun 20, 2025 at 3:19 PM Isabella Rossellini irossellini86@gmail.com wrote:

...

Dear Douglas,

Thank you for your kind help and advice! Good to know that a two-stage model is not needed for one subject. So I should not run the longitudinal preprocessing either, right?

Following your advice, I created this FSGD file below for subjid1 with 5 timepoints (0, 3, 6, 9, 12). Name of the imaging files: subjid1_base, subjid1_3mo, subjid1_6mo, subjid1_9mo, subjid1_12mo

GroupDescriptorFile 1 Title Onesubject Class Class1 Variables TimePoint Input subjid1_base Class1 0 Input subjid1_3mo Class1 3 #Input subjid1_6mo Class1 6 Input subjid1_9mo Class1 9 Input subjid1_12mo Class1 12

Does this look okay to you?

You also wrote that I should "remove the mean of the time points from the time point number number". (The mean is 6 in our case) I wonder why? Does this mean that the TimePoint variables should be -6 -3 0 3 6 instead? (and not 0 3 6 9 12)

GroupDescriptorFile 1 Title Onesubject Class Class1 Variables TimePoint Input subjid1_base Class1 -6 Input subjid1_3mo Class1 -3 #Input subjid1_6mo Class1 0 Input subjid1_9mo Class1 3 Input subjid1_12mo Class1 6

Sincerely, Isa Rosselini

On Thu, Apr 17, 2025 at 5:37 PM Isabella Rossellini irossellini86@gmail.com wrote:

...

Dear Douglas,

I am sorry for the late response. I did not see your message up until now. I would like to see how the brain changes for a patient with anxiety who got better gradually (again, we have 5-time points (0, 3, 6, 9, 12). And then do the same for another participant who did not get better up until 12 months. That is the reason I would like to look at patients one by one.

I ran the mri_glmfit analysis but I received an error message:

ERROR: DOF=0

Is this because I tried to run it on one participant?

Sincerely, Isa Rosselini

Date: Mon, 3 Feb 2025 09:20:40 -0500 From: "Douglas N. Greve" dgreve@mgh.harvard.edu Subject: Re: [Freesurfer] longitudinal analysis on one subject To: freesurfer@nmr.mgh.harvard.edu Message-ID: ebc3545a-9335-4b43-b7ae-da3c0b1d0bdb@mgh.harvard.edu Content-Type: text/plain; charset="utf-8"

The glm can be run on this data in theory. With only 5 time points, you probably won't have much power though. What are you trying to test?

On Sat, Feb 1, 2025 at 11:56 PM Isabella Rossellini irossellini86@gmail.com wrote:

...

Hello Freesurfer users/experts,

I am working with Freesurfer 7.4.1 on a MacStudio (macOS 15.2) and I would like to ask your opinion about the following.

Basically, we had an anxiety treatment study and I would like to analyze one subject who has 5-time points (0, 3, 6, 9, 12) with a longitudinal pipeline, two-stage model. Is it possible to run this on one person only and see the changes between these time points?

I have never done this before on one subject so I am not sure what the best approach would be to look at the changes. After running the longitudinal preprocessing (base, long), should I generate the tables of freesurfer parcellation/segmentation stats data (aparcstats2table, asegstats2table)? Or can I just follow the two stage model steps? Maybe glm only works on a group of subjects. https://secure-web.cisco.com/1oKKNLruyxLrTLstwDHVKJ7dOdikOj62Ii7jaQSAHbttB7p...

(We have 9 more subjects - 10 in total - so I was hoping to visualize the pattern of changes one by one, in each subject, since these patients showed different changes in anxiety based on self-reported questionnaires)

Sincerely, Isa Rosselini

---

Freesurfer mailing list -- freesurfer@nmr.mgh.harvard.edu To unsubscribe send an email to freesurfer-leave@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman3/lists/freesurfer@nmr.mgh.harvard.e...