Hi Sean,
symmetrized percent change is a measure across time. for two time points it is just 100*(tp2-tp1)/(0.5*(tp1+tp2)) (if they are a time unit (usually year) apart, else we divide by time difference)
You are basically fine. We need to distinguish between processing and post-processing. For processing it makes sense to have 3 time points in the base. You will get a subject template that initializes all time points. This enables you to to a 3time point analysis. It also reduces variability etc. If you run the stream with only two time points in the base for each interval independently you will basically only increase variability of your measures.
Of course in post-processing you can look at each interval independently. So looking at the difference in the first interval and in the second separately and then comparing them is totally fine. Also SPC is fine, but: as noted above in SPC we divide by the average . That average of course will change between the intervals and it might be better to divide by the average of all three time points instead (there is no script for that). I don't think it'll make a big difference, though. Alternatively you could look at the difference per time (rate).
On the cortex there is so much other variability (noise, non-linear registraton etc). You need to check if what you see is noise or significant. Of course it will go up and down in many regions, but that can all be noise. You should also take a look at the rate, because there you don't divide by the average.
And your final question: when looking at spc of all three time points it is NOT the same as viewing only tp1 to tp3. SPC of all three time points is: 100*slope/(temporal average) we fit a line into each subject data and compute the slope (this is the change per time: rate) and divide by the temporal average which is the measure at the mid time. This is assuming there is a linear change
Hope that helps. Best, Martin
On Jun 28, 2011, at 9:34 AM, Seán Froudist Walsh wrote:
Hi Martin and all,
I would like to ask you if by comparing tp1 thickness-spc with tp2 (and comparing tp2 with tp3) using the base brain between tp1,2 and 3 (altogether) I have messed up methodologically.
I was expecting a rise in cortical thickness in certain areas between tp1 and tp2 followed by a greater rise between tp2 and tp3. I did not expect however a host of other changes which were not seen when viewing spc of tp1, 2 and 3 together. Most of these effects seem to be basically mirror-effects e.g. if there is a rise between tp1 and tp2, there is a drop in thickness in the same area between tp2 and tp3 and vice versa. This happened in several subjects and in variable brain regions. I would like to know if this is because of some methodological flaw. For example, should I run the whole longitudinal stream 2 more times (once to compare tp1 and tp2, and the second time to compare tp2 to tp3)?
Also, when viewing the spc of tp1, tp2 and tp3 together are we just looking at the change between timepoints 1 and 3?
Many thanks for the help,
All the best,
Seán
On 27 June 2011 11:50, Seán Froudist Walsh froudiss@tcd.ie wrote: Hi Martin,
That worked! Thanks a lot,
Seán
On 23 June 2011 18:22, Martin Reuter mreuter@nmr.mgh.harvard.edu wrote: Hi Sean,
I think the --out... names need to be without the ending (mgh) so for example --out-avg=long23.thickness-avg
Maybe that'll fix it.
Best, Martin
On Thu, 2011-06-23 at 09:12 -0700, Seán Froudist Walsh wrote:
Dear FreeSurfers,
I am looking for a little help comparing longitudinal data. I have three timepoints on each patient and have been able to get individual spc-thickness maps using the long_mris_slopes command where my table included all three time points. I would however like to compare different timepoints with each other: tp3-tp2 and tp2-tp1.
I was hoping it would be as simple as specifying a new qdec table including (for example) only tp3 and tp2, and giving new output names using the following command
long_mris_slopes --qdec ./qdec/long.qdec.tableTP23.dat --meas thickness --hemi lh --do-avg --do-rate --do-pc1 --do-spc --do-stack --do-label --out-avg=long23.thickness-avg.mgh --out-rate=long23.thickness-rate.mgh --out-pc1=long23.thickness-pc1.mgh --out-spc=long23.thickness-spc.mgh --out-stack=long23.thickness-stack.mgh --out-label=long23.cortex --time scan --qcache fsaverage
but I get the following error:
mris_calc: Sorry, but I seem to have encountered an error. While making backup of internal data arrays, it seems that some of the backups already exist.
I'm not really sure what the internal data arrays are, and would greatly appreciate any help.
Many thanks in advance and all the best,
Seán
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at http://www.partners.org/complianceline . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail.