Hi Kasper,

generally the answer is yes, include all time point in the base, qc/edit and do whatever analysis you want on the final .long.base directories. This will guarantee a few nice things: 1. you will only need to do 1 pass through with QC and editing 2. information from other time points can help cleanup results for earlier time points (e.g. if 6 month image has motion, including year 1 and year 2 can help clean up surfaces there). 3. You have 1 consistent set of measurements, independent of what time points will actually be analyzed.

These argument usually overweight the disadvantages: 1. You need to wait until all data is there (in your case no disadvantage) 2. The longer you wait (in time) the further the base will move away from the initial time points. This is a disadvantage in your case, as 2 years could have some significant atrophy, and if you are only interested in test retest or 6 weeks (which is basically test-retest, unless you do some amazing magic there), you may be better off to run the test-retest component separately, to avoid getting an influence from future real atrophy cases. Usually this influence is limited in my tests, but no thorough analysis has been done on this yet.

So it depends on your study, on what you expect to happen, on you time for QC and editing (so also on your sample size). Etc. Also if you are looking for real effects in only 6 weeks, make sure you especially avoid/control biases such as hydration levels, or different motion levels, which could produce spurious effects:

hydration: http://www.ajnr.org/content/early/2015/09/17/ajnr.A4508 Head motion: http://www.sciencedirect.com/science/article/pii/S1053811914009975

Cheers, Martin

On 11/02/2015 04:44 AM, Kasper Jessen wrote:

Dear FreeSurfer group,

I have a question regarding timepoints and analysis. For example, lets say we have a data-set of 80 subjects with the following timepoints: baseline, 6 weeks, 6 months, 1 year and 2 years follow-up.

Two different studies on the same data needs to be performed. a) Study_1 with two timepoints (baseline and 6 weeks) b) Study_2 with all time timepoints (baseline, 6 weeks, 6 months, 1 year and 2 years follow-up).

As i understand, the more time points the better the [BASE] you will get. The BASE is going to be edited

Therefore, even though Study_1 only is going to be using two timepoints, do you still recommend to include all the 5 time points in creating the [BASE] even though only 2 time points will be used?

Best regards, Kasper Jessen

---

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