Contents
How To Set Up
How To Extract The Raw Data From Scanner
How To Inflate Brains Automatically
How To Inflate Brains Manually
How To Make Block Design Comparisons
How To DO Retinotopy
How To Draw Your Results
How To Manipulate ROIs and Annotations
How To Set Up
run nmrenv, and set
$SUBJECTS_DIR and $SUBJECT correctly.
(or in my case, run my script called
nmrdaniel)
My script
nmrdaniel is a
script that has the following lines:
---cut here--
#If you want the FS-Fast software etc, use this:
source ~inverse/freesurfer_alpha/nmr-std-env
setenv SUBJECTS_DIR /space/doodles/1/users/MTstudy/fMRI/subjects
setenv SUBJECT subj011
echo "All ready now. Subject value is:"
echo $SUBJECT
--cut here--
You must run either nmrenv or a script like nmrdaniel before running
any freesurfer or FS-FAST command.
generate the subject directory
structure:
mksubjdirs <subjname>
This command must be run for every subject.
Unpack the data from bourget:
browse-sessions
find the correct subject, possibly by
searching.
Ask it to start, and you will find
that instead
it will tell you the path for unpacking.
unpacksdcmdir -src
<source>
-targ <target> -scanonly
This outputs the type of scans used. Now
use
a file at $FREESURFER_HOME/numaris4-protocols.unpackcfg to correctly
make
a config file. The format is:
num folder format fileprefix
Here is an example config file.
1 scout bshort f
3 bold bshort f
4 bold bshort f
5 bold bshort f
6 bold bshort f
7 bold bshort f
8 bold bshort f
The number is acquisition number.
The folder is where that acquisition
is
stored (bold directory, scout directory etc)
The format is the form of the data
files.
The fileprefix is how the filenames
start.
Next run it again with your config file.
unpacksdcmdir -src
<source>
-targ <target> -cfg<configfile> -fsfast
Note: I have come across old data which
was
*.ima format. This can be unpacked using
unpackimadir2
which works just like unpacksdcmdir.
*** There is always a problem after
unpacking that the TR is not set correctly. Please fix it.
pico
$SUBJECTS_DIR/$SUBJECT/bold/seq.info
Goto the 3danat/ dir
Copy contents to mri/orig. For
example...
cp 3danal/* mri/orig
How To Inflate Brains Automatically
From the subject's directory, run
the following
recon-all -subjid
<subjname>
-all [-waitfor <filetowaitfor>] -notify <filetonotifydone>
(many ways to run this function. See
recon-all -help)
Or, do the faster technique, of running
each step of recon all separately.
I have a script called "reconSTEPS"
in my fMRI Tools section of my website which will help you do this.
To run it properly, you must make at least 2 copies of it, one for each
hemisphere, and edit them appropriately.
If you do it right, you can reconstruct the entire brain in less time,
since the hemispheres can be run in parallel.
Also, if you have more than one computer available, you can speed this
process up.
If you have more than one subject, you can reconstruct more than one of
them at a time using this technique.
The advantage of using reconSTEPS is that if anything fails you can
simply re-run the remaining steps instead of the
entire thing all over again. You can also go into manual
reconstruction, which is explained below.
To check results:
check_subject <subjname>
At this point, it should be good. But if
not,
the following steps may help.
How To
Inflate Brains Manually
A. Try a different structural.
cp 3danal/003/* mri/orig
B. If there is a filling error, you may
be
missing a good seed point in corpus collusum, right or left hemisphere
or
pons. Find them, then fill them based on your findings.
tkmedit <subjname> T1
Click on the points, and you can find
x,y,z.
mri_fill -C <x>
<y> <z>-rh <x> <y> <z> -lh <x>
<y> <z> -P
<x> <y> <z> $SUBJECTS_DIR/$SUBJECT/mri/wm/
$SUBJECTS_DIR/$SUBJECT/mri/filled
In the filled/ directory, make sure a
“surf” directory exists. If not, make one.
In that same directory, if
mri_tesselate is there. If not, run:
mri_tesselate_filled 255
../surf/lh.orig
mri_tesselate_filled 127
../surf/rh.orig
inflate_subject-rh<subjectname>
inflate_subject-lh<subjectname>
fix_subject<subjectname>
morph_subject<subjectname>
make_final_surfaces<subjectname>
parcellate_subject<subjectname>
map_all_labels
<subjectname>
C. Try using csurf:
Goto subject's directory.
csurf
Choose subject.
Subject-tools->Edit Segmentation.
Click on a region where two parts of
brain
are connected that should not be.
In tksurfer, TOOLS->save point.
In mkedit window, TOOLS-> goto
point.
Edit->voxels.
Middle mouse adds, right mouse
deletes.
Check slice before and after as well. Remove the unwanted connection.
File-> save main volume as->save
as
default
Surfer window,
Subject-tools->Process volume
Subject-tools->Create surface
Subject-tools->Fix surface topology
Subject-tools->Make final surface
Subject-tools->Create surface
Subject-tools->register surfaces
After all this...
To check results:
check_subject <subjname>
How To Make Block Design Comparisons
You will need paradigm files. They list
each TR based time point, tab, and then a condition number. For
instance:
0 0
2 0
4 1
6 1
8 0
...
This file is then saved as whatever you
feel like, but the same file name is used for all functional runs.
Tradition has "mtstudy.par". It is then placed in bold/003 where
003 is an example functional run. All runs get one.
You need to generate a file that lists
the subject name in the subject directory. Redundant here, but other
directory structures might care. This file must be called subjectname.
pico
$SUBJECTS_DIR/$SUBJECT/subjectname
And it simply contains the name of your
subject,
for instance
bert
This file is stored in the subject dir.
You need a totally separate study
directory,
from where you will make all your commands. In there, you need the
session
dir and session id files.
This file can be called anything, but by
my
custom use
sessdir
This file has one line, which is the
complete
path of the session directory, meaning the directory right underneath
the
one that has the name of the subject.
The session id file has the name of
the
subject on one line. The file is called
sessid.
Now you can start. From your study
directory,
run commands.
Make talairach xform files for later
registration
mktalxfm-sess -sf sessid -df sessdir
Motion correct
mc-sess -fstem f -fmcstem fmc -sf
sessid -df sessdir [-rlf runlistfile]
Spatial smooth and intensity normalize:
(note that 8 is arbitrary)
spatialsmooth-sess -i fmc -o fmcsm8
-fwhm
8 -sf sessid -df sessdir
Motion correction:
plot-twf-sess -ymax 5.0 -mc
-funcstem fmcsm8 -sf sessid -df sessdir
If motion > 2.5mm, time points can
be
disregarded. If excessive, drop that file.
Now make an tpexclude.txt file on the
subject's functional raw directories.
It contains one column of return
delimited numbers which represent each TR time point that should be
excluded.
emacs tpexclude.txt &
Make the analysis here. You can ignore
the
tpexclude if all your motion is ok. You can add the -runlistfile option
if
you want to.
mkanalysis-sess -analysis mtstudy
-paradigm mtstudy.par -tpexclude tpexclude.txt -blocked -motioncor
-inorm -sf sessid -df sessdir
OR
mkanalysis-sess.new -analysis
mtstudy -TR 2.5 -paradigm mtstudy.par -designtype blocked -funcstem
fmcsm8 -nskip 4 -polyfit 2 -gammafit 2.25 1.25 -timewindow 53
-nconditions 4 -runlistfile <runlistfilename>
-sf sessid -df sessdir
Edit the analysis setup file from the
studyDir:
emacs mtstudy/analysis.cfg &
Make average session level data...
selxavg-sess -analysis mtstudy -sf
sessid -df sessdir
Make contrasts...
mkcontrast-sess -analysis mtstudy
-contrast <contrastname>-a <#> -a<#>
-c <#>-c <#> -sf sessid -df sessdir
Now calculate them...
stxgrinder-sess -analysis mtstudy
-contrast <contrastname> -sf sessid -df sessdir
This can be simplified if there are more
than
one contrast, they can be run all at once:
stxgrinder-sess -analysis mtstudy
-all-sf sessid -df sessdir -space tal
stxgrinder-sess -analysis mtstudy
-all-sf sessid -df sessdir -space native
stxgrinder-sess -analysis mtstudy
-all-sf sessid -df sessdir -space sph
Now autoregister:
autoreg-sess -sf sessid -df sessdir
manual register after
tkregister-sess -sf sessid -df
sessdir
(there is also a thing called
tkregister2)
Resample (res is size of pixel:
1,2,4,8mm):
func2tal-sess -analysis mtstudy
-res 2
-sf sessid -df sessdir
func2sph-sess -analysis mtstudy -sf
sessid
-df sessdir
To make group averages (fixed effects), first
set up a group with
isxavg-fe-sess -analysis mtstudy
-group <groupname> -sf sessid -df sessdir -space <tal
or sph or native>
Then run the averaging with:
stxgrinder2-sess -analysis mtstudy
-s
<groupname> -space <tal or sph or native> -contrast
<contrastname>[-all] -sf sessid -df sessdir
To compare thresholded contrasts together,
use this guy...
mergecontrasts-sess -mergecontrast
<new_contrast_name>
-conjunction <and,or,andor>
-analysis
mtstudy -contrast <contrast1> <thresh>
<pos,neg,abs>
<1 if complement, 0 else> [
-contrast <contrast2>
<thresh>
<pos,neg,abs> <1 if complement, 0 else> ] [
-contrast
<contrast3>
<thresh> <pos,neg,abs> <1 if complement, 0 else>] [
...]
-map <t,minsig,iminsig,sig>
-space
<tal,sph,native>
-sf sessid -df sessdir
How To DO Retinotopy
For each session, create retinotopy paradigm files in each of the the
run directories. These paradigm files are different than an
event-related or block paradigm (which list which stimulus was
presented when). A retinotopy paradigm file has information about
whether the run was an eccentricy or polar stimulus and in what
direction the stimulus was presented.
For polar, the direction is indicates whether the spoke was traveling
clockwise or counter-clockwise. For eccen, the direction is
indicates
whether the ring was expanding or contracting. The definition of
what's positive and what's negative is arbitrary, though it must be
consistent. If you don't have both directions, just use positive.
For example, if there were four runs (001, 002, 003, 004), two eccen
and two polar, both in the positive and negative directions. Create
a file (eg, rtopy.par) in each run. Assuming run 001 was eccen in the
negative direction, then the rtopy.par for run 001 would look like:
--------cut here --------
stimtype eccen
direction neg
--------cut here --------
or
-------cut here------
stimtype polar
direction pos
--------cut here----
If needed, create a run-list file with the all retinotopy runs
(regardless of whether it was eccen or polar).
You need to generate a file that lists the subject name in the subject
directory. Redundant here, but other directory structures might care.
This file must be called subjectname.
pico
$SUBJECTS_DIR/$SUBJECT/subjectname
And it simply contains the name of your
subject,
for instance
bert
This file is stored in the subject dir.
You need a totally separate study
directory,
from where you will make all your commands. In there, you need the
session
dir and session id files.
This file can be called anything, but by
my
custom use
sessdir
This file has one line, which is the
complete
path of the session directory, meaning the directory right underneath
the
one that has the name of the subject.
The session id file has the name of
the
subject on one line. The file is called
sessid.
Now you can start. From your study
directory,
run commands.
Make talairach xform files for later
registration
mktalxfm-sess -sf sessid -df sessdir
Motion correct
mc-sess -fstem f -fmcstem fmc -sf
sessid -df sessdir [-rlf runlistfile]
Spatial smooth and intensity normalize:
(note that 5 is arbitrary)
spatialsmooth-sess -i fmc -o fmcsm5
-fwhm 5 -sf sessid -df sessdir
Motion correction:
plot-twf-sess -ymax 5.0 -mc
-funcstem fmcsm8 -sf sessid -df sessdir
If motion > 2.5mm, time points can
be
disregarded. If excessive, drop that file.
Now make an tpexclude.txt file on the
subject's functional raw directories.
It contains one column of return
delimited numbers which represent each TR time point that should be
excluded.
emacs tpexclude.txt &
Create the analysis:
mkanalysis-sess.new -analysis
<analysis_name> -TR
<TRvalue> -designtype
retinotopy -paradigm rtopy.par -funcstem
fmcsm5 -ncylcles <number
of cycles> [-runlistfile
<filename>]
Run the analysis:
sfa-sess -sf sessid -df sessdir
-analysis <analysis_name>
This will create <subjectname>/bold/<analysis_name> in
which will be:
1. volume h-offset (the mean function intensity)
2. eccen - eccentricity folder (with h volume)
3. polar - polar folder (with h volume)
The task-related activation can be viewed as with any other contrast
in a non-retinotopy application. Just use eccen or polar as the
contrast and specify h as the map.
The task-related activation for eccen can be viewed with:
sliceview-sess -sf sessid -analysis <analysis_name> -c eccen -map h -slice mos
The task-related activation for polar can be viewed with:
sliceview-sess -sf sessid -analysis <analysis_name> -c polar -map h -slice mos
Run paint:
paint-sess -analysis <analysis_name> -sf sessid
This will create map-imag-lh.w (the imaginary or sine part muliplied
by the log10(sig)), map-real-lh.w (the real or cosine part muliplied
by the log10(sig)), and sig-0-lh.w (the task related activation) for
both the polar and eccen. It will also create a directory called
fieldsign in bold/<analysis_name> in which fieldsign-lh (or -rh)
along with
fieldsignmask-lh will be found.
To view the field-sign on the flattend patch (occip.patch.flat):
surf-sess -sf sessid -analysis <analysis_name> -retinotopy fieldsign -flat
To view the eccen on the flattend patch (occip.patch.flat):
surf-sess -sf sessid -analysis <analysis_name> -retinotopy eccen -flat
To view the polar on the flattend patch (occip.patch.flat):
surf-sess -sf sessid -analysis <analysis_name> -retinotopy polar -flat
To view the task-related polar activation on the flattend patch
surf-sess -sf sessid -analysis <analysis_name> -c polar -flat
How To Draw Your Results
Paint (for surfing)
paint-sess-new -analysis mtstudy
-contrast <contrastname>-map <type> -sf
sessid -df sessdir
View individual with... (map types are
t,sig,minsig
and iminsig)
sliceview-sess -slice <#>
-analysis mtstudy -contrast <contrastname>-map <type>
-sf
sessid -df sessdir [-space tal]
surf-sess-new -analysis mtstudy
-contrast <contrastname>-map t-sf sessid -df sessdir
-hemi lh
surf-sess-new -analysis mtstudy
-contrast<contrastname>-map t-sf sessid -df sessdir -hemi
rh
To view volumes:
tkmedit-sess -analysis mtstudy
-contrast <contrastname> -map t -sf sessid -df sessdir [-space
tal]
View group data in surface form using these
(note you pick a subject's brain surface to draw upon...:
paint-sess-new -analysis mtstudy -s
<groupname> -contrast <contrastname>-map
<sig,t,minsig,iminsig> -d . -space <sph,tal,native>
-isxavg <fixed, random> -subject <subjectname>
surf-sess-new -analysis mtstudy -s
<groupname> -contrast <contrastname> -map
<sig,t,minsig,iminsig>-d . -space sph -isxavg <fixed
or random> -subject <subjectname> -hemi <lh
or rh> -fthresh <num> -fmid <num>
The simplest command for drawing freesurfer surfaces is:
tksurfer
But lets say you have a functional overlay? No problem. Use:
qsurfer
How To Manipulate ROIs and Annotations
To
make a set of annotations for the entire
brain, use this command.
parcellate_subject <subjname>
To convert the annotations into
labels:
mri_annotation2label --subject
<subjname> --hemi <hemi> --outdir <dir, starting from
subjdir> --table $FREESURFER_HOME/Simple_surface_labels2002.txt
--surf <which surface you want>
To copy labels from one brain to
another:
($lab=label name, $sub=subject number
mri_label2label --srclabel
~/MTstudy/fMRI/subjects/subj008/label/$lab
\
--srcsubject subj008 --trgsubject
subj0$sub
\
--trglabel
~/MTstudy/fMRI/subjects/subj0$sub/label/$lab \
--regmethod surface --hemi rh
--trghemi
rh --srchemi rh
Also, there is an optional flag --annotation aparc.a2005s to use the
new annotation type.
Matti's tool for batch label2label (includes a smoothing step to
make
sure the entire brain is covered):
mne_morph_labels