Hi Martin,
For 3., I don't think "time" was accidentally passed as a measure, as the command was just e.g. long_mris_slopes --qdec ./qdec/FPS/long.qdec.table.dat --meas area --hemi lh --do-pc1 --do-stack --do-label --stack-rate --time day --qcache fsaverage --sd $SUBJECTS_DIR therefore it is probably a bug; not important I guess, as long as that file can just be ignored.
If the way I ran the QDEC analyses is correct, then I will just ask in a separate email about that MC correction error. By MC I meant the Monte Carlo null-z correction included in QDEC (sorry for having been cryptical about it).
Many thanks indeed for your help.
Tudor
On 20 June 2014 16:08, Martin Reuter mreuter@nmr.mgh.harvard.edu wrote:
HI Tudor,
you could ask this as a qdec question, because it does not matter if you analyze thickness across groups or in your case thickness change. It is one measure per subject. It is essentially a cross sectional design (stage 1 takes the longitudinal aspect out of it). Your design looks OK. I don't know what MC correction is, and have not seen that error.
- yes, it maps your results to fsaverage space (where things need to be
in order for qdec to analyze/compare them). It also smoothes the data at different levels. This is necessary. 2. just ignore the days column (this can probably be done in Qdec, as I said, I am not very familiar with it) 3. probably you passed as a measure? Or maybe there is a bug when specifying --time? not sure 4. see 3, maybe there is a bug, or your table could be corrupted (e.g. column separators too many or to little so that for a single subject/time point columns are shifted left or right?
Best, Martin
On Jun 18, 2014, at 2:31 PM, Tudor Popescu tudor3@gmail.com wrote:
Many thanks for replying Martin. I now created 3 separate qdec tables (attached), for each of the 3 pairs of groups that I will analyse separaly in qdec. Since you said that doesn't affect stage1 (as the percentage change from tp1 to tp2 etc are the same regardless of number of groups), I did not run stage1 again for each pair but just used the original qdec table (i.e. "FPS", with all 3 groups included).
Is it conceptually correct that in QDEC, for each of the 3 designs, I included
- first *both *groups, and looked at the contrast showing whether
there is a difference in the e.g. thickness-pc1 between groups, i.e. whether anything changed *differently *from pre to post
- then *each *group separately, and looked at the cotnrast showing
whether thickness-pc1 is different from zero, i.e. whether anything changed from pre to post within that group?
Please let me know if I should post this as a separate QDEC question (although I get a feeling it's related to the longitudinal pre-processing), but when running these analyses for all measures (thickness and area for now), for area-pc1 analyses I get this error when wanting to start the MC correction@0.05: "Completed loading of analyzed data. fwhm.dat: 8.654965, rounded to 9. ERROR: CSDread(): could not open /usr/local/freesurfer/average/mult-comp-cor/fsaverage/rh/cortex/fwhm9/abs/th13/mc-z.csd". In case it's relevant, for the thickness analyses (whose correction worked fine), the fwhm.dat was "10.190419, rounded to 11".
A couple of other things:
- I suspect it is the "--qcache fsaverage" flag, included in the
long_mris_slopes example, that takes the bulk of the time for this command to run. Is it necessary to include this flag every single I run stage1 (for thickness lh&rh, area lh&rh etc)? 2. I see that in the cross.qdec.table.dat that gets created with the "qdec --table" command, there is still a "day" column, which is constant at "3" for everyone (as the average between scan on day 1 and scan on day 5). Consequently, I still have "day" appearing in my list of QDEC regressors, which makes little sense as this is now the second (cross-sectional) stage. Is this a problem, or can I just ignore the temporal (day) factor? 3. After those stage1 commands, I get files such as --time.fwhm10.mgh created in the subjects_dir folder - I didn't find information about them anywhere which makes me think they are perhaps created in error? 4. I think I found the source of the error that I mentioned previously, the full text of which was
*long_mris_slopes --qdec ./qdec/FPS/long.qdec.table.dat --meas pial_lgi --hemi lh --do-pc1 --do-stack --do-label --stack-rate --time day --qcache fsaverage --sd $SUBJECTS_DIR Parsing the qdec table: ./qdec/FPS/long.qdec.table.dat Working in SUBJECTS_DIR: /vols/Scratch/tpopescu/learning/FSWorking on hemi: lhSubject-Template: F02Traceback (most recent call last): File "/usr/local/freesurfer/bin/long_mris_slopes", line 752, in <module> times = [float(entry[varidx]) for entry in tplist]ValueError: invalid literal for float(): F* Line 752 of your long_mris_slopes code suggested it has to do with the time-dimension column (in my case called "day"). Although in the tutorial it only says that columns fsid and fsid-base need to come first and second (and not that the time column needs to be third), it turns out it does need to be third - in my case it was fourth, and when I changed it to third, the commands worked. What's strange is that earlier when the commands worked for me, the order of the columns was still the old one, i.e. time column was the fourth.
I apologise for the many emails and thank you again for your invaluable help.
Tudor
On 17 June 2014 10:29, Martin Reuter mreuter@nmr.mgh.harvard.edu wrote:
Hi Tudor,
the first stage stays the same (it creates a tickness-rate file for each subject in the base directory). The second stage compares these across groups.
'pial_lgi' should work. I don't know what the error means (never seen this before). You need to make sure that you have the files lh.pial_lgi and rh.pial_lgi in all the .long.base subject dirs (for all time points). I wonder if maybe one of these files is missing or maybe it is corrupted. Can you see from the output if it processes some subjects/timepoints and then stops?
Best, Martin
On Jun 17, 2014, at 1:04 AM, Tudor Popescu tudor3@gmail.com wrote:
Hi Martin,
Thanks, I asked the QDEC-related question in a separate email, and it seems it might be possible to do (stage#2) comparisons 2 group at a time (leaving one group out). Would the *first *stage of the two-stage model, i.e. long_mris_slopes, have to change if the *second *stage (QDEC) is run as 3 separate analyses, or is the command from the tutorial http://surfer.nmr.mgh.harvard.edu/fswiki/LongitudinalTwoStageModel still valid?
Also, after running the 'long' step with -localGI, I then ran stage#1 (with qcache) for lGI: *long_mris_slopes --qdec ./qdec/long.qdec.table.dat --meas pial_lgi --hemi lh --do-avg --do-rate --do-pc1 --do-spc --do-stack --do-label --stack-rate --time day --qcache fsaverage --sd $SUBJECTS_DIR* but every time received some crytpical compiler error such as "ValueError: invalid literal for float(): F". I replaced pial_lgi with local_GI with no effect. Right now I only have the rates of change computed for thickness and area but not for lGI.
Thanks! Tudor
On 15 June 2014 21:12, Martin Reuter mreuter@nmr.mgh.harvard.edu wrote:
HI Tudor, most of your questions are qdec or statistical questions, please post them to the Freesurfer list (with a different subject line) and maybe there is someone who can answer them. Lot's of people don't look at this mail, because it has 'Longitudinal" in it, but how many groups can be done in Qdec is a completely separate question.
About the longitudinal aspect. If you do the 2-stage model, then you have thickness change as your dependent variable (instead of thickness in a cross sectional analysis). You can look at :
- independent of group, where is thickness change different from zero.
(or for each group separately).
- You can compare groups, e.g. ask where is the change in thickness
larger/smaller in one group vs the one of the other groups.
For more complex modeling, I would recommend to include a biostatistician and run linear mixed effects models. The tools for that are available in FS53. Including thickness as a co-variate may be problematic. It may become tricky to interpret results (and there can very well be a coreallation with thickness change). If thickness changes as a percent, then you'll find a correlation with rate (as the same percent change in a thicker region is a larger rate in mm/time). If thickness changes more like an absolute (e.g. 0.2 mm/year) than in a thicker region the percent change will be smaller than in a thinner region.
Best, Martin
On Jun 15, 2014, at 3:56 PM, Tudor Popescu tudor3@gmail.com wrote:
Thanks!
I'm a bit confused as to why designs with 4 or 6 groups can be analysed with QDEC (as per this https://surfer.nmr.mgh.harvard.edu/fswiki/FsgdExamples FSGD examples page) whereas designs with 3 groups cannot be. It seems to me that an even number of groups is QDEC-able while an odd number isn't, but is there any workaround so that I can still use QDEC for my analyses?
Since I'm guessing I might have to use mri_glmfit after all, I created a FSGD file (attached), with 3 groups and 9 variables (age + 8 behavioural measures indexing performance during and after the learning). I have subjects of both genders, but am not interested in gender effects - I just want to include gender in the model so as to control for the gender imbalance and "model out" variance caused by it. I'm guessing, though, that if in the contrasts, 1s and -1s are attributed based on group membership rather than gender, this should accomplish the purpose? Same thing goes for age, although it being continuous rather than categorical, it's easier conceptually.
As I am not assuming different slopes for the different crossings of group and (age+behavioural predictors), I have built the FSGD as a DOSS.
Some issues:
- Some subjects have missing values for certain behavioural regressors.
Since I know there can't be empty cells in a FSGD (and thus in the design matrix), is it correct that my options are limited to either dropping the subject or filling in the missing cells with some "expected value", e.g. the mean value of that variable for the rest of the sample?
- It says here https://surfer.nmr.mgh.harvard.edu/fswiki/Fsgdf4G0Vthat
"NregressorsDOSS = Nclasses + Nvariables", which suggests you need regressors for group memberships, although such regressors are not present in any of the example FSGDs (and I've not included them in my file)
- Because there are no membership regressors, and also because the
"time dimension" (in my case: PRE or POST) is missing from this FSGD file, I am not sure how the contrast vectors should be built, given the longitudinal aspect of the data. So the research questions that I'd like these contrasts to ask are:
- Where in the brain does thickness change significantly from PRE to
POST, for the 3 different groups (pairwise)?
- Where in the brain are these changes correlated with (each of) the
behavioural measures? - For each subject, how is the change in thickness predicted by the PRE thickness value in that area, *or *by the behavioural regressors? In other words, can individual brain differences before training predict differences in learning abilities indexed after training?
Many thanks for your help! Tudor
On 13 June 2014 17:27, Martin Reuter mreuter@nmr.mgh.harvard.edu wrote:
Hi Tudor,
- There is a new version of freeview, I think you can download it
somewhere. The tutorial also gives the tk surfer command line, which will work. But the visualization is not necessary. 2.probably the qdec table has errors, eg different line breaks? Can you split it into blocks to see which block it does not load? 3. To look at change you'd look at rate or percent change. Avg is just the average thickness of each subject across time. I never really used qdec, but probably you cannot do three groups.
Best Martin
Sent via my smartphone, please excuse brevity.
-------- Original message -------- From: Tudor Popescu Date:06/12/2014 10:00 PM (GMT+01:00) To: free surfer Subject: Re: [Freesurfer] Longitudinal design
Hi Martin
There were in fact some problems with the lGI values, as it wasn't clear at which of the 3 preprocessing steps (cross, base, long) the --pial_lgi flag had to be added. I asked about that in a different email, and meanwhile continued with the analysis for the existing measures (area and thickness). Issues I encountered:
- After using the tutorial command freeview -f
subj/surf/lh.pial:overlay=subj/surf/lh.long.thickness-avg.fwhm15.mgh:overlay_threshold=0,3.5:overlay=subj/surf/lh.long.thickness-stack.mgh:annot=subj/label/lh.aparc.annot:annot_outline=1 --timecourse --colorscale I received errors about the option 'timecourse' and the sub-option flag 'annot_outline' being unrecongised. Is there a more recent version of freeview? I'm using FS 5.3, which I thought was the latest. Either way, I guess that this particular visualisation is not essential to the analysis.
- When I load my cross.qdec.table.dat (attached) into QDEC, "Number of
subjects" in the Subjects tab shows as 67, even though I have 72 subjects, all of which appear correctly entered on each line of the cross.qdec.table.dat !
- If I understood correctly, if one wants to see how thickness changed
across time points and across the different groups, i.e. test the interaction, then "long.thickness-avg" has to be selected as Measure in QDEC. After doing this, and choosing group as a discrete factor, the mri_glmfit --y command fails with the errors "ERROR: QdecGlmDesign::GenerateContrasts: factor 1 must have 2 levels". I guess this is because I have 3 groups and not 2. The question is, can I somehow do the 3 pairwise comparisons between my 3 groups in QDEC rather than at the command line (mri_glmfit) ?
Many thanks! Tudor
On 7 June 2014 23:24, Martin Reuter mreuter@nmr.mgh.harvard.edu wrote:
Hi Tudor. For the mri_concat command at the end, does it work when you run it directly? It should not. Are the input files there. Is the dimension correct (you can run mri info on them). Since they are just surface maps, just like thickness, their dimensions should agree (eith each other and with thickness, and number of vertices of surfaces).
Best Martin
Sent via my smartphone, please excuse brevity.
-------- Original message -------- From: Tudor Popescu Date:06/07/2014 9:55 PM (GMT+01:00) To: Freesurfer support list Subject: Re: [Freesurfer] Longitudinal design
I should add that the commands for thickness, i.e. long_mris_slopes --qdec ./qdec/long.qdec.table.dat --meas thickness --hemi ?h --do-avg --do-rate --do-pc1 --do-spc --do-stack --do-label --time day --qcache fsaverage --sd $SUBJECTS_DIR finished without error, and for some reason, so did the surface area commands, when I ran them a second time. Also, I realised that for lGI, the flag for --meas was pial_lgi and not just lgi. However, the command long_mris_slopes --qdec ./qdec/long.qdec.table.dat --meas pial_lgi --hemi lh --do-avg --do-rate --do-pc1 --do-spc --do-stack --do-label --time day --qcache fsaverage --sd $SUBJECTS_DIR resulted in this error, similar for both hemispheres: SUBJECT F02 Stacking Within-Subject Maps mri_concat /vols/Scratch/tpopescu/learning/FS/F02_d1.long.F02/surf/rh.pial_lgi /vols/Scratch/tpopescu/learning/FS/F02_d5.long.F02/surf/rh.pial_lgi --o /vols/Scratch/tpopescu/learning/FS/F02/surf/rh.long.pial_lgi-stack.mgh ERROR 1 : mri_concat stacking did not work?
Thanks again for your help!
On 6 June 2014 17:45, Tudor Popescu tudor3@gmail.com wrote:
Hi Martin, thanks a lot for your answers. The slopes command is giving me errors though, see below. I also attach the long.qdec.table.dat that I'm using. I should point out that those ?h.long.cortex.label files mentioned in the error messages are in fact at their expected locations, and appear to be fine (i.e. are not size 0)
In case anyone else can advice about question 1, I'd be really grateful!
*Command*
*Error*
long_mris_slopes --qdec ./qdec/long.qdec.table.dat --meas area --hemi lh --do-avg --do-rate --do-pc1 --do-spc --do-stack --do-label --time day --qcache fsaverage --sd $SUBJECTS_DIR
No such file or directory
mris_label_calc: could not open label file /vols/Scratch/tpopescu/learning/FS/F02/label/lh.long.cortex.label
Invalid argument
mris_label_calc: mris_label_calc.cpp:127: int main(int, char**): Assertion `l2 != __null' failed.
long_mris_slopes --qdec ./qdec/long.qdec.table.dat --meas area --hemi rh --do-avg --do-rate --do-pc1 --do-spc --do-stack --do-label --time day --qcache fsaverage --sd $SUBJECTS_DIR
mris_calc: could not open label file /vols/Scratch/tpopescu/learning/FS/F23/label/rh.long.cortex.label
Invalid argument
mris_calc: could not read label file /vols/Scratch/tpopescu/learning/FS/F23/label/rh.long.cortex.label
Invalid argument
long_mris_slopes --qdec ./qdec/long.qdec.table.dat --meas lGI --hemi lh --do-avg --do-rate --do-pc1 --do-spc --do-stack --do-label --time day --qcache fsaverage --sd $SUBJECTS_DIR
mris_label_calc: mris_label_calc.cpp:127: int main(int, char**): Assertion `l2 != __null' failed.
SUBJECT F02 Intersecting Within-Subject Cortex Label
cp /vols/Scratch/tpopescu/learning/FS/F02_d1.long.F02/label/lh.cortex.label /vols/Scratch/tpopescu/learning/FS/F02/label/lh.long.cortex.label
mris_label_calc intersect /vols/Scratch/tpopescu/learning/FS/F02_d5.long.F02/label/lh.cortex.label /vols/Scratch/tpopescu/learning/FS/F02/label/lh.long.cortex.label /vols/Scratch/tpopescu/learning/FS/F02/label/lh.long.cortex.label
ERROR -6 : mris_label_calc intersect did not work?
long_mris_slopes --qdec ./qdec/long.qdec.table.dat --meas lGI --hemi rh --do-avg --do-rate --do-pc1 --do-spc --do-stack --do-label --time day --qcache fsaverage --sd $SUBJECTS_DIR
No such file or directory
mris_label_calc: could not open label file /vols/Scratch/tpopescu/learning/FS/F02/label/rh.long.cortex.label
Invalid argument
mris_label_calc: mris_label_calc.cpp:127: int main(int, char**): Assertion `l2 != __null' failed.
On 5 June 2014 16:00, Martin Reuter mreuter@nmr.mgh.harvard.edu wrote:
> Hi Tudor, > > 1. not sure, this is a more general question for people more > familiar with surfaces and edits > > 2. nothing is necessary, however, if you have regions where you find > difficulties in some subjects, I'd look at that region in all subjects > (e.g. you mention the posterior end). Usually if surfaces on the base are > good, the long should be high quality also. > > 3. looks good (I'd do space separated, also some columns merged in > your email and fsid-base for subject3 is wrong). > > 4. using the two-stage model you could do a pairwise analysis > between your groups in the second stage, so linear mixed effects model is > not necessary. In the two stage model, just think of it as a cross > sectional analysis of e.g. thickness change, instead of thickness. > > 5. yes, is sutiable. You'd have to run it several times, once for > each of the measures you want to analyze. You can drop some of: > > --do-avg --do-rate --do-pc1 --do-spc --do-stack > > and only keep e.g. the rate or the pc1 (depending on what you want > to analyze). You can also use --stack-rate to generate a stack of the rate > maps. > > Best, Martin > > > On 06/04/2014 01:19 PM, Tudor Popescu wrote: > > Hi Martin, > > I reran everything from scratch and this time the recons ended > without error. Few more questions: > > 1) Is the brainmask.mgz vs wm.mgz problem that I described in my > previous email (surfaces only starting to show rather late at the posterior > end of some subjects) something that I need to correct (e.g., by editing > wm.mgz), or is it fine as it is? > 2) Given that I visually inspected all bases, Is it necessary to > inspect *all *slices for the longitudinal overlaps (vols+surfs) of > all subjects, or is it enough to just check the central coronal slice for > each subject? I ask because it's not just *opening *freeview with > all those layers that is very slow, but also navigating between slices > 3) The tutorial > http://freesurfer.net/fswiki/FsTutorial/LongitudinalTutorialcovers > the creation of long.qdec.table.dat for a single-group scenario; I have 3 > groups, with scans on the first and final (fifth) day of training, and was > wondering whether the following table looks correct > > *fsid* *fsid-base* *group* *day* > *ageDemeaned * *gender* *score* subj1_day1 subj1 treatmentA 1 > 1.34 f 109 subj1_day5 subj1 treatmentA 5 1.34 f 109 subj2_day1 > subj2 treatmentB 1 -5.85 m 102 subj2_day5 subj2 treatmentB 5 -5.85 > m 102 subj3_day1 subj2 control 1 3.06 f 98 subj3_day5 subj2 > control 5 3.06 f 98 > 4) Is this mixed design (3 groups and 2 time points, with both > effect of group and of time being relevant) suitably analysed with only a > simple 2 stage model, or is a linear mixed effects model necessary? Also, > how are the stats different given there are 3 groups and not the usual 2 > than most FS tutorials are based on? > 5) Is the default long_mris_slopes command (which I understand does > preliminary analyses, before going to QDEC) mentioned in the tutorial, i.e. > > long_mris_slopes --qdec ./qdec/long.qdec.table.dat --meas thickness --hemi lh --do-avg --do-rate --do-pc1 --do-spc --do-stack --do-label --time years --qcache fsaverage --sd $SUBJECTS_DIR > > suitable for my particular case? I will also look at surface area > and lGI, so I suspect it's at least "--meas area" and "--meas lGI" that I > need to add to the above command? > > Thanks so much! > Tudor > > > > > On 3 June 2014 02:02, Martin Reuter mreuter@nmr.mgh.harvard.edu > wrote: > >> HI Tudor, >> >> you can run the -long processes in parallel. Something else is >> wrong. Check if you have enough disk space. Re-run the long from scratch >> (to make sure everything is generated new). >> >> Best, Martin >> >> On Jun 2, 2014, at 5:34 PM, Tudor Popescu tudor3@gmail.com >> wrote: >> >> Hi Bruce >> >> I suspected this might be the case, since assessing 3D surfaces >> from a succession of 2D slices can be tricky, which is why I was wondering >> whether there's a better way to do these visual inspections. I checked >> those apparent artefacts in the other 2 planes (axial, saggital), but did >> not find it helpful in deciding whether the surfs was as they should or not. >> >> Something that I think is more likely to be a real problem (i.e., >> more than just a viewing plane disorientation), is that for many subjects, >> posterior slices only start to be followed by the surfaces quite late, e.g. >> from coronal slice 80 onward. This seems to be because those slices only >> appear in the brainmask.mgz, but not also in the wm.mgz (see screenshots >> attached). >> >> Also, after having run the cross and base steps, and ended up >> with folders of the type subj1_scanA, subj1_scanB etc, for the "long" step >> I launched simultaneously (for all subjects&scans) commands of the type >> recon-all -long subj1_scanA subj1 -all >> recon-all -long subj1_scanB subj1 -all >> however these resulted in two types of error messages: >> * >> mghRead(/vols/Scratch/tpopescu/learning/FS/subj1_scanA.long.subj1/mri/aseg.mgz, >> -1): read error* >> (those particular aseg.mgz files had size 0kB), and >> * (standard_in) 2: Error: comparison in expression* >> I initially had disk space issues, but after making space I >> relaunched the same commands with the -no-isrunning flag, but with the >> same results. Is it the case that the long commands for the different scans >> of a subject have to be launched sequentially rather than in parallel, i.e. >> the command for scanB should be run after the command for scanA has >> finished? Otherwise, what could the problem be? >> >> Thanks again for your help. >> >> Tudor >> >> >> >> On 1 June 2014 15:17, Bruce Fischl fischl@nmr.mgh.harvard.edu >> wrote: >> >>> Hi Tudor >>> >>> are you assessing the thickness visually? If so, you are probably >>> being misled by the orientation of the surface w.r.t. the viewing plane. >>> Same fo the apparent bubbles. You need to look in a different orientation >>> >>> cheers >>> Bruce >>> >>> >>> On Sun, 1 Jun 2014, Tudor Popescu wrote: >>> >>> Many thanks Martin and Nick! The rendering option was already >>>> checked, but >>>> the freeview inspection commands were still very very slow, so I >>>> used >>>> commands of the following form, which as far as I understand >>>> should be >>>> equivalent: >>>> tkmedit subjID norm.mgz -surfs >>>> For several subjects (maybe 20 out of the total of 72), I noticed >>>> that the >>>> surfaces don\t seem to quite follow the GM/WM demarcation line as >>>> expected - >>>> for instance, by having portions of white surface (yellow line, >>>> see attached >>>> screenshots) surrounded by pial surface (red line), or by having >>>> very large >>>> cortical thickness in some parts of the brain. Is it possible >>>> that these >>>> apparent artefacts make sense "in context" (by looking at the >>>> adjacent >>>> slides), or is it the case that I need to go back and do white >>>> surface >>>> correction, and then redo all 3 steps of the longitudinal process? >>>> >>>> Thanks again for your help! >>>> Tudor >>>> >>>> >>>> On 29 May 2014 17:39, Nick Schmansky nicks@nmr.mgh.harvard.edu >>>> wrote: >>>> Tudor, >>>> >>>> in your virtual machine, make sure the 3D rendering option, >>>> or >>>> 'use >>>> hardware rendering' (or something like that) is enabled. >>>> this >>>> would be on >>>> the Windows VM config side of things. otherwise it will do >>>> software >>>> rendering which is painfully slow. >>>> >>>> Nick >>>> >>>> >>>> > Hi Tudor, >>>> > >>>> > I don't think there is a way to speed things up. >>>> > Let me know if you find a case where the template is >>>> blurry or >>>> has >>>> > ghosts. It should not happen, but if it does it indicates >>>> a >>>> bad >>>> > registratration, you'd have to run the mri_robust_template >>>> command with >>>> > different parameters manually then. >>>> > >>>> > Best, Martin >>>> > >>>> > On 05/27/2014 06:13 PM, Tudor Popescu wrote: >>>> >> >>>> >> Hi Martin, >>>> >> Wasn't sure whether you'd seen my reply below, look >>>> forward >>>> to hear >>>> >> back your thoughts, thanks! >>>> >> Tudor >>>> >> >>>> >> On 25 May 2014 21:40, "Tudor Popescu" <tudor3@gmail.com >>>> >> mailto:tudor3@gmail.com> wrote: >>>> >> >>>> >> Thanks very much Martin and Bruce. I guess I'd misread the >>>> Wiki >>>> >> (my own fault, not the text's), and am glad to hear that >>>> the >>>> >> longitudinal pipeline is in fact perfectly suitable for my >>>> needs >>>> >> here. >>>> >> >>>> >> Having run the first 2 steps (cross and base), I'm a bit >>>> unclear >>>> >> how the output so far has to be manually inspected. It >>>> says in >>>> the >>>> >> tutorial >>>> >> < >>>> http://freesurfer.net/fswiki/FsTutorial/LongitudinalTutorial%3E >>>> >> that you should load each subject's base volume + surfs in >>>> >> freeview and then "move back and forth a few slices". >>>> However, >>>> >> even just loading each base in this manner takes ~1 min on >>>> my >>>> PC >>>> >> (CoreDuo, 4GB, Ubuntu Virtualbox in Windows 7), and then >>>> moving >>>> >> with PgUp/PgDn between all coronal slices (starting from >>>> the >>>> >> default slice=128, going all the way posterior and then >>>> all the >>>> >> way anterior) is excruciatingly slow. All of this would >>>> have to >>>> be >>>> >> repeated for all my 72 subjects - is there any way to >>>> optimise >>>> >> this manual inspection? >>>> >> >>>> >> Also, if the surfs turn out to not follow the volume >>>> correctly, >>>> >> presumably the thing to do is white surface correction + >>>> >> re-running recon. But what should one do if, due to an >>>> erroneous >>>> >> averaging between timepoints, you see blurs/ghosts in your >>>> base >>>> >> template? >>>> >> >>>> >> Many thanks! >>>> >> Tudor >>>> >> >>>> >> >>>> >> On 9 May 2014 21:33, Martin Reuter < >>>> mreuter@nmr.mgh.harvard.edu >>>> >> mailto:mreuter@nmr.mgh.harvard.edu> wrote: >>>> >> >>>> >> Hi Tudor, >>>> >> >>>> >> the longitudinal pipeline in FS is actually one of the >>>> best >>>> on >>>> >> the planet as far as I know :-). If there is any >>>> contradictory >>>> >> information on the wiki, can you point me to that so I >>>> can >>>> see >>>> >> what causes the misconception. Really: compared to >>>> independent >>>> >> processing, it significantly increases sensitivity. >>>> Also we >>>> >> have designed it to be unbiased with respect to a >>>> single >>>> time >>>> >> point or directionality. It is quite mature by now. >>>> >> >>>> >> You should definitely use the longitudinal pipeline >>>> for the >>>> >> analysis of your data. Now to your questions >>>> >> >>>> >> 1. QDEC: I am not too familiar with qdec. You can >>>> definitely >>>> >> try the 2-stage approach described on the wiki. There >>>> you >>>> >> first compute a measure of change (e.g. hippocampal >>>> volume >>>> >> change during your week) and then compare that measure >>>> across >>>> >> groups similar to a cross sectional volume/thickness >>>> analysis. >>>> >> You can also use our tools to run a linear mixed >>>> effects >>>> model >>>> >> if you want to do that (it is more involved and >>>> requires >>>> you >>>> >> to use matlab tools). In your case, you probably have 2 >>>> time >>>> >> points for all subjects and the time distance is >>>> probably >>>> the >>>> >> same for all subjects, so the 2-stage approach should >>>> be >>>> fine. >>>> >> >>>> >> 2. The image processing is done via the longitudinal >>>> pipeline >>>> >> (three steps: cross, base, long), to prepare the data >>>> look >>>> at >>>> >> the description of the 2-stage model >>>> >> http://freesurfer.net/fswiki/LongitudinalTwoStageModel >>>> >> and also the longitudinal tutorial >>>> >> >>>> http://freesurfer.net/fswiki/FsTutorial/LongitudinalTutorial >>>> >> >>>> >> 3. At the recon all level in FS you get (after the 3 >>>> steps) >>>> >> measurement for all time points. So you would compare >>>> those >>>> >> results across time in the stats. >>>> >> >>>> >> Hope that helps, Martin >>>> >> >>>> >> >>>> >> On 05/08/2014 08:14 AM, Tudor Popescu wrote: >>>> >>> Sorry for the repeat, wasn't sure whether this was >>>> received >>>> >>> the first time. >>>> >>> Tudor >>>> >>> >>>> >>> >>>> >>> On 6 May 2014 19:55, Tudor Popescu <tudor3@gmail.com >>>> >>> mailto:tudor3@gmail.com> wrote: >>>> >>> >>>> >>> Dear FS list, >>>> >>> >>>> >>> I have structural data from a learning study >>>> >>> (pre&post-training scans, with 3 groups). >>>> Although the >>>> >>> training was only one week, I'm guessing from an >>>> analysis >>>> >>> point of view, this still qualifies as >>>> longitudinal. I >>>> >>> want to check for >>>> >>> >>>> >>> * the main within-subjects effect of time point >>>> >>> (pre&post) >>>> >>> * the main between-subjects effect of group >>>> (treatment >>>> >>> A, treatment B, control), >>>> >>> * the time x group interaction >>>> >>> >>>> >>> I intend to look at thickness, surface area, >>>> volume, >>>> and >>>> >>> lGI. >>>> >>> >>>> >>> I read on the wiki >>>> >>> >>>> http://surfer.nmr.mgh.harvard.edu/fswiki/LongitudinalProcessing >>>> >>> that FS is currently not optimal for longitudinal >>>> >>> analyses. I intend my FreeSurfer analysis to >>>> supplement a >>>> >>> VBM analysis done in FSL. In case it is in fact a >>>> good >>>> >>> idea to do this, my questions (not covered in the >>>> >>> 'longitudinal' wiki page) are: >>>> >>> >>>> >>> 1) Can QDEC be used for such an analysis, and if >>>> so, >>>> what >>>> >>> would be different as compared to a >>>> cross-sectional >>>> (no >>>> >>> temporal/within factor) study? >>>> >>> >>>> >>> 2) Also, is the pre-processing stage any >>>> different? >>>> >>> >>>> >>> 3) In FSL, for longitudinal designs you do stats >>>> on >>>> >>> images obtained as the difference between >>>> consecutive >>>> >>> time points. Does this have to be done in >>>> FreeSurfer >>>> as >>>> >>> well, and if so, is it done at the recon-all >>>> level or >>>> >>> only at the stats (QDEC) level? >>>> >>> >>>> >>> Thanks! >>>> >>> >>>> >>> Tudor >>>> >>> >>>> >>> >>>> >>> >>>> >>> >>>> >>> >>>> >>> _______________________________________________ >>>> >>> Freesurfer mailing list >>>> >>> Freesurfer@nmr.mgh.harvard.edu >>>> >>> mailto:Freesurfer@nmr.mgh.harvard.edu >>>> >>> >>>> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer >>>> >> >>>> >> -- >>>> >> Martin Reuter, Ph.D. >>>> >> >>>> >> Instructor in Neurology >>>> >> Harvard Medical School >>>> >> Assistant in Neuroscience >>>> >> Dept. of Radiology, Massachusetts General Hospital >>>> >> Dept. of Neurology, Massachusetts General Hospital >>>> >> Research Affiliate >>>> >> Computer Science and Artificial Intelligence Lab, >>>> >> Dept. of Electrical Engineering and Computer >>>> Science, >>>> >> Massachusetts Institute of Technology >>>> >> >>>> >> A.A.Martinos Center for Biomedical Imaging >>>> >> 149 Thirteenth Street, Suite 2301 >>>> >> Charlestown, MA 02129 >>>> >> >>>> >> Phone:+1-617-724-5652 tel:%2B1-617-724-5652 >>>> >> Email: >>>> >> mreuter@nmr.mgh.harvard.edu >>>> >> mailto:mreuter@nmr.mgh.harvard.edu >>>> >> reuter@mit.edu mailto:reuter@mit.edu >>>> >> Web :http://reuter.mit.edu >>>> >> >>>> >> >>>> >> _______________________________________________ >>>> >> Freesurfer mailing list >>>> >> Freesurfer@nmr.mgh.harvard.edu >>>> >> mailto: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. >>>> >> >>>> >> >>>> >> >>>> >> >>>> >> _______________________________________________ >>>> >> Freesurfer mailing list >>>> >> Freesurfer@nmr.mgh.harvard.edu >>>> >> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer >>>> > >>>> > -- >>>> > Martin Reuter, Ph.D. >>>> > >>>> > Instructor in Neurology >>>> > Harvard Medical School >>>> > Assistant in Neuroscience >>>> > Dept. of Radiology, Massachusetts General Hospital >>>> > Dept. of Neurology, Massachusetts General Hospital >>>> > Research Affiliate >>>> > Computer Science and Artificial Intelligence Lab, >>>> > Dept. of Electrical Engineering and Computer Science, >>>> > Massachusetts Institute of Technology >>>> > >>>> > A.A.Martinos Center for Biomedical Imaging >>>> > 149 Thirteenth Street, Suite 2301 >>>> > Charlestown, MA 02129 >>>> > >>>> > Phone: +1-617-724-5652 >>>> > Email: >>>> > mreuter@nmr.mgh.harvard.edu >>>> > reuter@mit.edu >>>> > Web : http://reuter.mit.edu >>>> > >>>> > _______________________________________________ >>>> > 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 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. >>> >>> >> <02-06-2014, 22.17.45.png><02-06-2014, 22.17.55.png><only LH >> posterior.png><norm.png>_______________________________________________ >> >> >> Freesurfer mailing list >> Freesurfer@nmr.mgh.harvard.edu >> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer >> >> >> --------------------------------- >> Dr. Martin Reuter >> Assistant in Neuroscience - Massachusetts General Hospital >> Instructor in Neurology - Harvard Medical School >> MGH / HMS / MIT >> >> >> A.A.Martinos Center for Biomedical Imaging >> 149 Thirteenth Street, Suite 2301 >> Charlestown, MA 02129 >> >> Phone: +1-617-724-5652 >> Email: >> mreuter@nmr.mgh.harvard.edu >> reuter@mit.edu >> Web : http://reuter.mit.edu >> >> >> _______________________________________________ >> 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. >> >> > > > _______________________________________________ > Freesurfer mailing listFreesurfer@nmr.mgh.harvard.eduhttps://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer > > > -- > Martin Reuter, Ph.D. > > Instructor in Neurology > Harvard Medical School > Assistant in Neuroscience > Dept. of Radiology, Massachusetts General Hospital > Dept. of Neurology, Massachusetts General Hospital > Research Affiliate > Computer Science and Artificial Intelligence Lab, > Dept. of Electrical Engineering and Computer Science, > Massachusetts Institute of Technology > > A.A.Martinos Center for Biomedical Imaging > 149 Thirteenth Street, Suite 2301 > Charlestown, MA 02129 > > Phone: +1-617-724-5652 > Email: > mreuter@nmr.mgh.harvard.edu > reuter@mit.edu > Web : http://reuter.mit.edu > > > _______________________________________________ > 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. > >
<g3v9.fsgd>
Dr. Martin Reuter Assistant in Neuroscience - Massachusetts General Hospital Instructor in Neurology - Harvard Medical School MGH / HMS / MIT
A.A.Martinos Center for Biomedical Imaging 149 Thirteenth Street, Suite 2301 Charlestown, MA 02129
Phone: +1-617-724-5652 Email: mreuter@nmr.mgh.harvard.edu reuter@mit.edu Web : http://reuter.mit.edu
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Dr. Martin Reuter Assistant in Neuroscience - Massachusetts General Hospital Instructor in Neurology - Harvard Medical School MGH / HMS / MIT
A.A.Martinos Center for Biomedical Imaging 149 Thirteenth Street, Suite 2301 Charlestown, MA 02129
Phone: +1-617-724-5652 Email: mreuter@nmr.mgh.harvard.edu reuter@mit.edu Web : http://reuter.mit.edu
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.
<FP_cross.qdec.table.dat><FPS_cross.qdec.table.dat> <FS_cross.qdec.table.dat><PS_cross.qdec.table.dat> _______________________________________________
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Dr. Martin Reuter Assistant in Neuroscience - Massachusetts General Hospital Instructor in Neurology - Harvard Medical School MGH / HMS / MIT
A.A.Martinos Center for Biomedical Imaging 149 Thirteenth Street, Suite 2301 Charlestown, MA 02129
Phone: +1-617-724-5652 Email: mreuter@nmr.mgh.harvard.edu reuter@mit.edu Web : http://reuter.mit.edu
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.