Hi Ken,
you should always look at the output from the …long.base directories, those are
the final outputs. Looks to me that you are looking at the cross sectional
produced surfaces from the first stage. Those will be
- in different spaces (as the heads are in different positions in the
Hi Marisa,
That should be possible.
You can run the long pipeline up to that step, replace the corresponding files
with your version and continue from there.
Best, Martin
> On 1. Oct 2020, at 10:32, Marisa Johanna Nordt wrote:
>
> External Email - Use Caution
> Hello,
>
Hi Marisa,
the -affine flag was only for some debugging and is currently not supported. It
may also not do what you think it does. It will remove scaling difference
between your time points. This makes sense if you have test-retest scans and
want to remove scanner calibration scaling
Hi Suzan,
(sending this also to the FS support list, where these questions fit best, but
sometimes I miss emails there).
If edits are necessary or not depends purely on the quality of your data. I.e.
old scanners, low contrast, motion artefacts etc.
It also depends on the number of
Hi Marisa,
usually a random time point is selected as an initial registration target. From
there we map all images (including the initial target) to a common mid space
and create the first template there, then we register to that template and
iterate. This speeds things up, but is
Hi Marisa,
the main goal of robust_template is to remove rigid motion between scans. So
the first question is, do you really need (want) affine?
The next question is, if this still occurs in FS 7.1.1 I remember dimly that I
worked on this some years ago (basically allow a larger threshold, as
Hi Ken,
Hmm, not sure. Does it give a reason why it cannot be loaded? Could be read
permissions, could be corrupted.
Generally the commands look good. You could also try to check the subject SPC
on the base-template surface. Those files should also be there in the base dir.
Best, Martin
Hi Barbara,
I think you are asking to run FastSurfer and source FS7.1 binaries instead of
the recommended 6.0. for the surface processing part.
This may work, but it is not recommended. We only tested FastSurfer in
combination with 6.0. as described in the paper. So if you use 7.1 you are on
Or should I merge unsmoothed data (e.g.,
> lh.thickness.fwhm0.fsaverage.mgh) and then smooth once merged?
>
> Thanks again,
> Lara
>
>
> From: freesurfer-boun...@nmr.mgh.harvard.edu <
> freesurfer-boun...@nmr.mgh.harvard.edu> on behalf of Ma
Hi Guillaume,
not sure what is going on. It could be missing permissions or it could
be that the disk is full?
You could also try to drop the --do-label and see if that works.
Best, Martin
On Fri, 2020-02-14 at 10:12 +0100, Guillaume Carey wrote:
>
> Hello everyone,
> I’m currently
Hi,
looks like someone changed the default python to version 3, and you are
still using a script written for python 2. I think Andrew updated this
recently, so obtaining a newer version of the long_submit_jobs script
from dev may fix this.
Best, Martin
On Tue, 2020-02-11 at 18:03 +,
Hi Lara,
no, you should not use time point 1 as base.
You need to distinguish between the image processing part and the
statistics post processing analysis (LME). These are two separate
things.
For the image processing in the longitudinal pipeline, we create the
base (=subject template). The
Hi Swati,
- yes, you can treat the percent change maps (PC1) as the signal instead of the
thickness and do a regular analysis (so either multiple comparisons or sim
etc).
- and yes, I would always recommend to use the linear mixed effects model
instead as it is specifically designed for
Hi,
I have never tested compatibility with 5.1 and would recommend against it (5.1
was released almost 9 years ago). You can try it out on one subject and check
results. There would be no support for this however.
Best, Martin
> On 17. Dec 2019, at 20:34, Greve, Douglas N.,Ph.D.
>
Hi Jamie,
looks like your command is wrong (reverse).
Here are spaces:
tp1 original space
tp2 original space
base = tp1.long.base = tp2.long.base space
so the longitudinals and the base are in the same space (rigidly aligned and
resliced to that space).
The lta files map between tp1
; absolute volume for any comparative analysis.
>
> Thank you again. This message board is of incredible value!
>
> Best regards,
> David
>
>
>
>
>> ------
>>
>> Message: 1
>
Hi David,
I am not very optimistic:
5mm is too thick for FreeSurfer (recommendation is 1 up to 1.5). You
will certainly get something, but it can be very unreliable and
completely wrong. Especially longitudinally these thick slices will
induce large variance due to different head positioning
Hi Doug,
probably something went wrong during the editing. E.g maybe you edited
files in the cross sectional or base directories and copied/saved these
to the longitudinals or vice-verca. There should be no mis-alignments.
I would recommend this:
1. run one such subjects without edits through
Hi Alexandru,
not sure what you mean with "unmatched" error.
If this is still a problem, can you send details?
Best, Martin
On Wed, 2019-08-07 at 08:31 -0400, Alexandru Hanganu wrote:
> External Email - Use Caution
> Hello Martin,
>
> I was wondering - how solved the
Hi Seonjoo,
your base command is missing the "-all" so no processing is done in the
base, which is why the file is missing later..
Best, Martin
On Wed, 2019-08-07 at 17:27 +, Lee, Seonjoo wrote:
> External Email - Use Caution
> Dear Freesurfer,
> Could you help me to
Hi, To debug this, I would need the 4 input images and the command. Of course the error would need to also come up when using those images.Best MartinAm 14.08.2019 20:01 schrieb "Greve, Douglas N.,Ph.D." :I have not seen it before; I've cc'ed Martin in case he has seen it.
Have you checked the
Hi Sierra,
the long command should be
recon-all -long C002d C002.base -mprage …..
same way as you called it the first time when using the -all flag.
Best, Martin
> On 8. Jun 2019, at 20:28, Sierra Ann Jarvis wrote:
>
> External Email - Use Caution
>
>
> Hello
Hi Adam,
the longitudinal registration should not care, but the regular stream would
like to have the skull in the image. This will help for a couple steps (full
head talairach registration -> eTIV etc).
Best, Martin
> On 5. Jun 2019, at 03:57, Adam Martersteck wrote:
>
>
Hi Martin,
probably the reviewer does not mean anything related to FS. Rather,
there is distortion during image acquisition (e.g. gradient non-
linearities). The reviewer probably wants to know if gradient
unwarparing was done prior to running FS. That is why Matthew asked
what MRI data was
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 ?
Hi William,
you can use freesurfer's pctsurfcon to compute the gray/white contrast
on the longitudinal directories. This will produce a surface overlay
at: tp.long.base/surf/?h.w-g.pct.mgh
I don't think the long_mris_slopes script can deal with those files
directly, so you have two options:
-
Hi Maxime,
also how do the surfaces look on their respective time point images
from long? Do they look accurate there? Could be that there are
differences across the time points that cause the surfaces to differ
this much. Was the same hardware (scanner, coil) and same protocol used
for both
Hi Maxime,
please keep previous conversation in the email (so that it is easier to check
the original question). Also let’s keep this on the list.
I do not understand what you mean with “there is not strict alignment”. If you
open the norm.mgz from base, tp1.long.base and tp2.long.base they
Hi Maxime,
the longitudinal stream aligns images across time. This registration
can (and should) be checked by opening the base/mri/norm.mgz and the
norm.mgz from the tp.long.base (longitudinal) directories. They should
all align.
Your example looks like you might have openend the images from
Hi Arsenije,
the file name should be (with a starting “.” and no file extension):
$SUBJECTS_DIR/qdec/.Qdecrc
Best, Martin
> On 22. Feb 2019, at 00:22, Arsenije Subotic
> wrote:
>
> Dear Freesurfer experts,
>
> I am having issues choosing the spc option for my QDEC two stage
also note the file name starts with a “.”
.Qdecrc
> On 18. Mar 2019, at 21:16, Martin Reuter wrote:
>
> Hi Theo,
>
> the file should be :
> $SUBJECTS_DIR/qdec/.Qdecrc
>
> it is important that it is at that location and should not have an extension
> (no
Hi Theo,
the file should be :
$SUBJECTS_DIR/qdec/.Qdecrc
it is important that it is at that location and should not have an extension
(not .txt !)
Best, Martin
> On 18. Mar 2019, at 16:16, AKUDJEDU, THEOPHILUS
> wrote:
>
> External Email - Use Caution
> Dear All,
>
> I am
ut this effect would
> > be compensated by LME modeling ?
> >
> > Would there be proportion of early drop outs to respect in order to
> > compensate bias with LME ?
> >
> > Thanks,
> > Matthieu
> >
> >
> > Le ven. 19 oct. 2018 à 16:
Hi Elena,
FreeSurfer put inputs in RAS coordinates (during the conform step at
the very beginning), so it will take care of this. But only if the
correct information is available from the input image header. If that
header was corrupted earlier, it is impossilbe to know what coordinates
the
It says "Could not find file ..". I wonder which one it is looking for.
Did you try to completely delete this long subject dir and re-run from
scratch?
Do other time points with this base go through? If not, you could re-
run the base also.
Also when you open orig of all time points (from the
Hi Erik,
usually we parallelize the number of subjects, rather than the
individual pipelines. That gives us the best performance.
Best, Martin
On Thu, 2018-12-06 at 11:07 +, Erik O'Hanlon wrote:
> External Email - Use Caution
> Hi FS experts,
>
> I'm running a
Hi Kathleen,
if you map one image to the other, this introduces interpolation
artefacts, which will bias your analysis (any type of analysis, so this
is already a problem in the lesion segmentation part).
Technically you run longitudinal freesurfer on these images but
(similar to the lesion
Hi David,
there is no need to worry. Whenever we change the pipeline, there will be
consistent differences. Since we do not have ground truth thickness it is
unclear which method is “more right”. A comparison between cross and long does
not make sense here with respect to absolute values. It
Hi Martin,
ICV is head size. It is unlikely that anything except head growth in children
changes head size. I don’t know of drugs that do that ;-).
So it is best to assume that head size is fixed for adults. Usually results
should have less variance if you remove the noise from the ICV
Hi,
Looks like the input is a multi frame image. Should only be single frame.
Best Martin
> On 26. Oct 2018, at 13:52, Iglesias Gonzalez, Eugenio
> wrote:
>
> It seems like a problem with mri_robust_register.
> Martin, any ideas?
>
> --
> Juan Eugenio Iglesias
> Centre for Medical Image
gt; between both groups are still valid accounting for this bias ? Is LME
> method robust enough for compensating this kind of drop-out ?
>
> Best,
> Matthieu
>
>
> Le mer. 17 oct. 2018 à 18:33, Martin Reuter edu> a écrit :
> > Hi Matthieu,
> >
> > 1) surv
Hi Molly,
no one can really tell you that. My feeling is, that it is pretty safe,
given that you have already 4 time points in the base. The base will be
very stable and the main reason for the base is that it is an
independent space different from any single time point. Which remains
true.
al neuroimaging data ?
>
> Thanks in advance for helping.
>
> Best,
> Matthieu
>
> Le mer. 14 déc. 2016 à 22:14, Martin Reuter edu> a écrit :
> > Hi Matthieu,
> >
> > 1. yes, LME needs to be done first so that values can be sampled
> > from
Hi Rebecca,
I see. Generally, it should behave in the same way as cross sectional
processing:
If both aseg.mgz and aseg.auto.mgz exist, and they are different, then
the file aseg.manedit.mgz is created from the differences (and if an
aseg.manedit.mgz file also exists, then its contents are
Hi Rebecca,
aseg edits to subcortical structures need to be done in the -long
stage.
(see here: https://surfer.nmr.mgh.harvard.edu/fswiki/LongitudinalEdits
)
Best, Martin
On Thu, 2018-09-06 at 14:37 -0700, Rebecca Ray wrote:
> External Email - Use Caution
> HELP
> --
s of tp2 from tp1 for each subject
> in a tabular form.
>
> Best,
>
> Anwar
>
> On 2018-08-30, 3:22 AM, "Martin Reuter" .edu> wrote:
>
> Hi Anwar,
>
> what FS version are you using?
>
> Also what ha
Hi Xiaoyu,
please keep the previous conversation as inline citation below so I know what
we discussed before.
What I mean is to process every subject only once and not twice (with flipped
images, if that is what you did?).
Then run xhemi analysis on the subject-level. Invert the sign of
Hi Dani,
the regular FreeSurfer stream samples everything to 1mm isotropic. This is of
course problematic, if you have one group scanned at 1mm and the other at a
different resolution (as group differences could be affected by the different
resolutions). You would never know how much of what
Hi Anwar,
what FS version are you using?
Also what happens if you omit the --out-pc1 but keep the --do-pc1 ?
Also no output with that ending? It should write it to the default
location.
Best, Martin
On Wed, 2018-08-29 at 16:45 +, Shatil, Anwar Shahadat wrote:
> External Email
0-05152013
>
> Thank you!
> Valeria
>
> From: freesurfer-boun...@nmr.mgh.harvard.edu [freesurfer-bounces@nmr.
> mgh.harvard.edu] on behalf of Martin Reuter [mreu...@nmr.mgh.harvard.
> edu]
> Sent: Wednesday, August 29, 2018 12:14 PM
&
Hi Xiaoyu,
again (see also other email) you should not run left hemispheres on the
right. This could introduce a processing bias, especially when you
force all left to be healthy and right to be diseased (or vice-versa).
Instead run your images normally and then for the univariate , simply
Hi Valeria,
what version of FS are you using (also is it a local install or the
system install)?
Also try to re-type the command in the shell (don't run it from a
script) could be that there are invisible characters , like tab etc
from copy paste that mess things up.
Best, Martin
On Wed,
Hi Xiaoyu,
I don't think it makes sense to model the hemisphere each as a
different group. Do you mean different time point?
The problem is that left hemispheres are probably anatomically slightly
different than right hemispheres (on average in healthy people).
I would run every subject only
Hi Foad,
if all subjects have 2 time point and same time delta, you can compute the
difference between time points, divide by 2 (for yearly change) and run a
standard GLM analysis on that measure instead of thickness.
Or you can model this with linear mixed effects models (which makes sense
Hi Paul,
you will need aseg and maybe more in the -long runs anyway. But if you really
want to stop after norm
… -autorecon1 -gcareg -canorm
passing autorecon2 will do the whole block :
https://surfer.nmr.mgh.harvard.edu/fswiki/ReconAllTableStableV6.0
Hi Kaley,
the problem is,, when you re-run the base with different images, it will create
the median image at a different location. That means that all your edits will
be at the wrong location. I currently don’t see a way to fix this easily
without a lot of tweaking of the recon-all script.
Hi Paul,
Does this happen in the base or in the long step?
-bigventricle has never been tested within the long stream and this is bug that
needs to be fixed at some point (assuming the big ventricle is supported in the
next release )
We should put this on our known issues list.
Best,
Hi James,
I would not run upsampled data through the highres pipeline. (not even sure I
would run highres data through that).
So you should sample to 1mm resolution. To be fair you should also resample the
1mm image to a new position (to also introduce interpolation artefacts there).
You
Hi Ryan,
FreeSurfers Longitudinal pipeline assumes head size is relatively fixed. The
problem with growing heads is that the surfaces from the subject-template will
not fit well to later (or earlier) time points and that the algorithm can
potentially not recover. If surfaces look OK it means
Hi Laurel and Johannes,
Pediatric: if the time distance is small it should also work in paediatric
images. You can just test it on a couple cases and if too many edits would be
required, then simply use the cross sectional processed results.
There is no thorough analysis, but missingness is
Hi Kaley,
if you scroll down the log file you should find something like
new invocation of recon-all with the correct time stamp and v6 text. The old
log is kept and the new one is appended.
I am also not sure it is OK to re-run the base with 6.0. It could be that this
looks OK in single
Hi Arsenije,
I have never seen this. You could try:
- convert the norm.mgz to norm.nii.gz and back (keep a backup of the old one)
and see if that changes anything
- you can also re-run this case from scratch to see if maybe anything else is
wrong . Could have been an IO issue.
Best, Martin
Hi Kody,
scanner effects across vendor could be larger than disease effects. You can
attempt an analysis but you will have a tough time to publish it given these
differences and no real way to control for scanner effects. Maybe scan 5-10
people on all these scanners and show that scanner
Hi Samir,
you can first conform all inputs to 1mm isotropic using mri_convert --conform
input output
It may make sense to add a covariate in the statistics for the different
acquisition to control for a linear effect.
Best, Martin
> On 25. May 2018, at 08:29, Samir Rekic wrote:
>
>
>
>
> From: freesurfer-boun...@nmr.mgh.harvard.edu
> <mailto:freesurfer-boun...@nmr.mgh.harvard.edu>
> <mailto:freesurfer-boun...@nmr.mgh.harvard.edu>> on behalf of Martin Reuter
> mailto:mreu...@nmr.mgh.harvard.edu>>
> Sent: 15 May 2018 23:39
> To: Freesurfer suppor
Hi Tudor,
5.3 had some difficulties reporting image geometry differences even when they
were tiny. That was fixed in 6.0. However Bruce’s recommendation (to reslice
the second time point exactly as the first) should do the trick. Alternatively
(in order to avoid processing bias) you could
Hi Erik,
you should investigate why this is the case. Any difference in acquisition will
cause effects and it will be impossible to distinguish these from real effects.
Also you should never just resample one time point, as that will introduce
processing bias. I would advise to
- find out
Hi Tamir,
ICV is fixed across time (derived from the template image), so that data can be
compared directly, but for cross subject analysis you still need to adjust for
ICV of course.
Best, Martin
> On 15. May 2018, at 03:19, Tamir Eisenstein wrote:
>
>
>
> Hello
Hi Emmanuel,
I think this should be identical to the cross sectional situation, you map that
cluster (defined on FSAVERAGE) to each subject_time_point and then use
mri_segstats to get stats on that region. Probably others can help you with
what tools exactly you need for this (or probably
Hi Laura,
it could be that 9 years of severe aging or neundegenerative disease are too
far apart for the method to cover the distance. The base would look blurry or
show ghosting (in 2 time points the median is equal to the mean and the mean
will not look good with too much change). I would
Hi,Yes the longitudinal images after the long run are all aligned in the base space. You will find the map from each time point to the base in the base MRI transforms dir (maybe also in the long MRI transforms). Naming should make clear what reach one is.Best MartinAm 09.05.2018 21:11 schrieb
Hi Susanna,
yes, as you feared, you should rerun the first time point so that all time
points are run by the same version. Edits should be preserved (although maybe
more or probably less edits would be necessary).
And it is really a bad idea to compare across groups where the groups where
Hi Anna,
yes, long can still be edited, but usually this is not necessary. What kind of
edits did you need to do in the long?
You would need to call with the long flag, and passing the cross and base name.
So just like the regular processing of the -long step, only replace -all with
Hi Katie,
yes, you should check the base (subject-template) if further editing is
necessary there, so that surfaces look good on the base.
Best, Martin
> On 4. Apr 2018, at 15:54, Mckay, Katie Geraldine
> wrote:
>
> Hi,
> I have a question regarding the longitudinal
Hi Maria,
you can include all time points per subject in the longitudinal processing. For
the statistics I would add a time-varying co-variate with the head coil type.
Best, Martin
> On 13. Mar 2018, at 11:45, Maria Paternina Die wrote:
>
> Dear all,
>
> We want to run
Hi Lisa,
1. It all depends on what kind of analysis you want to do. You can run
aparcstats2table without the long_stats_slopes. Then you will get an
entry for each time point. If you run long_stats_slopes it will compute
the within-subject difference first, put the result into each base
s N Greve)
> 2. Re: Extracting the centroids from Desikan-Killiany atlas
>(zuxfoucault Wong)
> 3. Unsubscribe Mailing List (Duy Nguyen)
> 4. Unknown Label in following postprocessing with mne-python
>(Daniel van de Velden)
> 5. Re: Surface Err
# pbcmd = ...
with a hashtag.
you can also simplify the slurm command (remove the --mail flags and the
-o logfile location flags) your cluster partition may not be called
"work" either.
Best, Martin
Am 22.02.2018 um 16:43 schrieb Martin Reuter:
Hi Lea,
I wrote that script to
Hi Lea,
I wrote that script to simplify processing on our cluster which uses
qsub PBS for submission. I have never used SLURM so far.
There is two functions that you (or someone who knows this stuff) would
need to modify:
def submit
and
def wait_jobs
The submit procedure basically
foreach? surfreg --s $subject --t fsaverage_sym --lh --xhemi
foreach? end
Thanks,
Lanbo
On Tue, Jan 16, 2018 at 4:46 AM, Martin Reuter
<mreu...@nmr.mgh.harvard.edu
<mailto:mreu...@nmr.mgh.harvard.edu>> wrote:
Hi Lanbo,
Hi Sara,
the base (subject template) is pretty robust. Not sure what goes on that
it fails on 50% of your images. You should debug to see why it fails?
- look at the input images, e.g. the norm.mgz from both your time points
from the cross sectional processing
- check the log file for the
.
Secondly, how I can get the change rate after construct left -right
registration.
Thanks,
Lanbo
On Fri, Dec 22, 2017 at 5:40 AM, Martin Reuter
<mreu...@nmr.mgh.harvard.edu <mailto:mreu...@nmr.mgh.harvard.edu>> wrote:
Hi Lanbo,
you could look at longitudinal changes of th
Hi Lanbo,
you could look at longitudinal changes of the left-right difference in
volume per ROI. Or do you mean on the cortical thickness map (I have
never done that, but probably works similarly, construct left -right
registration, compute difference, then run the LME on that).
Best,
Hi Shane,
not really. You can of course run images through the stream. You will
get measurements, but you will not know how much of that is caused by
the scanner or by disease (or drug,etc). If you have 2 groups, you could
model the scanner as a co-var, but there could be a group scanner
Hi Tamara,
yes, it is in the aseg stats . You can extract that information for all
subjects using asegstats2table
Best, Martin
Am 23.11.2017 um 20:43 schrieb Tamara Tavares:
Hello,
Just a quick question regarding the computed intracranial volume. I
have ran Freesurfer version 5.1 and I
Hi Lara,
yes that is possible (and it also makes sense in early childhood as the
longitudinal stream assumes no (or only minimal) head growing).
Best, Martin
Am 14.11.2017 um 19:03 schrieb Lara Foland-Ross:
> Hello Freesurfer experts,
>
> I have two longitudinal datasets. In each dataset,
Hi Paul,
you can try to re-run from scratch (ensure there is enough space on the disk)
and if you can replicate the problem start dropping some of the flags (like the
parallel and the bigventricles etc, to find out if this is caused by one of the
flags.)
Anyway it looks more like an IO
Hi Lanbo,
the first command computes the slope of within-subject linear fits (the rate,
so the unit is mm/time). You use —generic time which means that the time is 1 2
3 … It is better to put the real time, e.g. in years for a longitudinal study.
the second command is a
Hi Garret,
sure, there is no need to run qcache at all. Assuming you do a thickness
analysis, you can directly use mris_preproc to register all thickness
maps from only the first time point to fsaverage and smooth them (there
you need to pass a list of only the first longidutinal time points:
Hi Chris,
thanks, can you send the images and the exact commands. We have had success in
the past, could be that you have a special case (e.g. maybe we never tested
registration across different resolutions or whatever?).
Thanks, Martin
https://gate.nmr.mgh.harvard.edu/filedrop2/
Hi Christoph,
1) I would use only intercept as random effect. I like to keep the model simple
and - as you say - there is only a max of 2 time points. If you like you can do
a model comparison (see wiki page) to compare the two models.
2) I would use the exact time in years (as float with
Hi Amy,
please send your questions to the freesurfer email list .
Try replacing ID with fsid , also you may need to fix the white spaces (replace
multiple spaces with a single space or tab).
Best, Martin
> On 9. Oct 2017, at 03:09, 田阿敏 <237416...@qq.com> wrote:
>
>
> Hi~
>
> I am Amy from
Hi Michelle,
it should work like this. I don't know if it was tested with the -3T and
-mprage flags. You can try to omit these flags (for testing only).
Also the SB017 has been processed cross sectionally before, right?
For debugging, you can also look at the individual steps in the base
the timepoints. Is
there a command to perform this operation?
Also, will the "-qdec-long" flag replace the "--s" flag because the
qdec table will contain the names of the fsid?
Best,
Paul
On Wed, Sep 27, 2017 at 11:38 AM, Martin Reuter
<mreu...@nmr.mgh.h
Hi Paul,
for the statistical analysis take a look at the available options here:
https://surfer.nmr.mgh.harvard.edu/fswiki/LongitudinalStatistics
for asegstats2table you can use the --qdec-long flag with an appropriate
text file containing the columns fsid and fsid-base . This will convince
Hi Caleb,
is this in a single subject or a group result? In a single subject
almost anything is expected. E.g. small motion can cause thickness
changes , so can hydration levels.
Best, Martin
Am 21.09.2017 um 16:53 schrieb Huang, Caleb (NIH/NINDS) [F]:
Hello,
I'm tracking atrophy in
Yes,
see also my answers from 08 and 11 of September on the list.
Best, Martin
Am 17.09.2017 um 19:42 schrieb Martijn Steenwijk:
The longitudinal pipeline will not overcome (contrast) differences due
to scanner differences. If you register the scans to eachother you
will see that there
Hi Lisa,
not sure where the problem comes from. But generally one would not do
the qcache on the base. Qcache maps results to fsaverage and nobody
should ever look at the base results except for debugging purposes. The
base is the within-subject template and it is used to initialize the
below I meant you could run it only on the first two time points. Maybe
there is already enough power to see effects.
Best, Martin
Am 08.09.2017 um 09:13 schrieb Martin Reuter:
Hi Michelle,
when scanners (or even scanner hardware, like head coil or software
upgrades etc) change
Hi Michelle,
when scanners (or even scanner hardware, like head coil or software
upgrades etc) change in a longitudinal study you will have difficulties
finding out whether the changes you see are real anatomical changes or
scanner effects. This is true for any software pipeline (longitudinal
1 - 100 of 816 matches
Mail list logo