External Email - Use Caution
Dear all,
I'm a beginner in using the longitudinal processing pipeline (as well as statistical analysis) and it would be great to get some insights or hints to analyze my data.
I have a dataset consisting of 11 subjects each acquired at 7 different time points with an isotropic resolution of 1 and 0.8 mm at 3T using a 64-channel head coil. Using that dataset I want to investigate short term differences in e.g. cortical thickness with the goal to assess the degree of biological variance during that time period.
I have plotted the mean cortical thickness of each time point of every subject (using lme_timePlot and lme_lowessPlot) showing a somewhat random distribution across time and from my perspective fairly high standard deviation. I wanted to have a look at the individual percent change by overlaying the symmetric percent change on fsaverage, but wasn't quite sure of the scale. Is it in percent? So in case I set the scale bar between 1 and 5, the color relates to 1 to 5 percent?
What other ways would make sense to have a look at? I definitely cannot compare groups, as there is just one. The days and time of acquistion are rather randomly choosen, so I potentially cannot use either as a covariate.
Best, Falk
...............................................
[klinikum_logo_schmal_e-mail_blau] University Clinic for Neurology
Otto-von-Guericke-university Magdeburg Medical faculty Leipziger Str. 44 39120 Magdeburg
Phone +49-391-6117-512
falk.luesebrink@med.ovgu.demailto:falk.luesebrink@med.ovgu.de http://www.kneu.ovgu.de/kneu.html
Hi Falk,
yes, the output of long_mris_slopes and long_stats_slopes is in percent (100 * rate / value_of_fit_at_mid_time).
Also running 1mm data is different (as you know :-) from .8 so maybe you would analyze both separately, e.g. creating one base on 1mm time points and another on the .8mm ?
If you re-scan in short time interals (and if these are young and healthy, then even for mid to long intervals), you would not expect anatomical aging effects. Then the variance is probably mainly acquisition noise (e.g. induced by motion etc, see eg https://www.ncbi.nlm.nih.gov/pubmed/25498430) plus some processing noise (different surface placement etc). Maybe there are also hydration effects (see https://www.ncbi.nlm.nih.gov/pubmed/26381562 ).
Best, Martin
On Thu, 2019-05-09 at 09:51 +0000, falk.luesebrink@med.ovgu.de wrote:
External Email - Use CautionDear all,
I’m a beginner in using the longitudinal processing pipeline (as well as statistical analysis) and it would be great to get some insights or hints to analyze my data.
I have a dataset consisting of 11 subjects each acquired at 7 different time points with an isotropic resolution of 1 and 0.8 mm at 3T using a 64-channel head coil. Using that dataset I want to investigate short term differences in e.g. cortical thickness with the goal to assess the degree of biological variance during that time period.
I have plotted the mean cortical thickness of each time point of every subject (using lme_timePlot and lme_lowessPlot) showing a somewhat random distribution across time and from my perspective fairly high standard deviation. I wanted to have a look at the individual percent change by overlaying the symmetric percent change on fsaverage, but wasn’t quite sure of the scale. Is it in percent? So in case I set the scale bar between 1 and 5, the color relates to 1 to 5 percent?
What other ways would make sense to have a look at? I definitely cannot compare groups, as there is just one. The days and time of acquistion are rather randomly choosen, so I potentially cannot use either as a covariate.
Best, Falk
...............................................
University Clinic for Neurology
Otto-von-Guericke-university Magdeburg Medical faculty Leipziger Str. 44 39120 Magdeburg
Phone +49-391-6117-512
falk.luesebrink@med.ovgu.de http://www.kneu.ovgu.de/kneu.html
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
External Email - Use Caution
Hi Martin,
thanks for your quick reply and the clarification regarding the output.
I have processed the 1 and 0.8 mm separately, so one base for each subject and resolution - no mixing has been conducted! :)
Actually, I want to look into changes due to hydration and other biological effects, e.g. due to hormone status. However, this is just a preliminary assessment of the variance in general (and getting to know the processing pipeline) as no meta data was acquired to correlate the measures against.
Best, Falk
-----Ursprüngliche Nachricht----- Von: freesurfer-bounces@nmr.mgh.harvard.edu freesurfer-bounces@nmr.mgh.harvard.edu Im Auftrag von Martin Reuter Gesendet: Montag, 13. Mai 2019 12:56 An: Freesurfer support list freesurfer@nmr.mgh.harvard.edu Betreff: Re: [Freesurfer] Longitudinal processing
Hi Falk,
yes, the output of long_mris_slopes and long_stats_slopes is in percent (100 * rate / value_of_fit_at_mid_time).
Also running 1mm data is different (as you know :-) from .8 so maybe you would analyze both separately, e.g. creating one base on 1mm time points and another on the .8mm ?
If you re-scan in short time interals (and if these are young and healthy, then even for mid to long intervals), you would not expect anatomical aging effects. Then the variance is probably mainly acquisition noise (e.g. induced by motion etc, see eg https://www.ncbi.nlm.nih.gov/pubmed/25498430) plus some processing noise (different surface placement etc). Maybe there are also hydration effects (see https://www.ncbi.nlm.nih.gov/pubmed/26381562 ).
Best, Martin
On Thu, 2019-05-09 at 09:51 +0000, falk.luesebrink@med.ovgu.de wrote:
External Email - Use CautionDear all,
I’m a beginner in using the longitudinal processing pipeline (as well as statistical analysis) and it would be great to get some insights or hints to analyze my data.
I have a dataset consisting of 11 subjects each acquired at 7 different time points with an isotropic resolution of 1 and 0.8 mm at 3T using a 64-channel head coil. Using that dataset I want to investigate short term differences in e.g. cortical thickness with the goal to assess the degree of biological variance during that time period.
I have plotted the mean cortical thickness of each time point of every subject (using lme_timePlot and lme_lowessPlot) showing a somewhat random distribution across time and from my perspective fairly high standard deviation. I wanted to have a look at the individual percent change by overlaying the symmetric percent change on fsaverage, but wasn’t quite sure of the scale. Is it in percent? So in case I set the scale bar between 1 and 5, the color relates to 1 to 5 percent?
What other ways would make sense to have a look at? I definitely cannot compare groups, as there is just one. The days and time of acquistion are rather randomly choosen, so I potentially cannot use either as a covariate.
Best, Falk
...............................................
University Clinic for Neurology
Otto-von-Guericke-university Magdeburg Medical faculty Leipziger Str. 44 39120 Magdeburg
Phone +49-391-6117-512
falk.luesebrink@med.ovgu.de http://www.kneu.ovgu.de/kneu.html
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
_______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
freesurfer@nmr.mgh.harvard.edu
-
falk.luesebrink＠med.ovgu.de -
Martin Reuter