MGH Analysis

This information is very old...most machines where this stuff were installed are gone...still, it is a lot of information, so I'd hate to just delete it when it might be useful to someone, somewhere.  Dianne Patterson 10/25/2006

This page describes Event Related Analysis with the New MGH program which only runs on Linux (and is still problemamatic. Perhaps you really want the Afni_Event page which describes a processing stream we have used successfully for years involving afni and mgh)

You can do analysis of individuals without talairaching their data, but then you cannot do group analyses (Fixed or Random Effects). You can talairach the data and do group analysis. With a little extra work, you can do both. The first several steps in the analysis are the same irrespective of your talairaching decision.

No Talairaching

Creating the Parameter File
  • For the Tutorial Data, par.txt files have been provided.
  • Purpose: The paradigm file is analagous to parts of the waver 1D file for afni and the model specification in SPM. It describes the timing of control and stimulus conditions throughout the functional run. You will create one paradigm file for each run and place a copy of the paradigm file into the directory with that run's .bshort files. The paradigm file is a text file containing two columns of numbers. The first column contains the stimulus onset time, in seconds, relative to the time at which the first stored image of the run was collected. The second column represents the condition code for each stimulus. By convention, the code 0 represents the baseline or fixation condition. The other conditions are to be numbered sequentially starting with 1. Give the file a name that describes your study, e.g. motorstudy.para. The .para extension is not required, but helps to identify the file as a paradigm file.

    For the event related paradigms, the time (column one) should be the time of stimulus onset and need not be an integer multiple of the TR. For example, in the paradigm file below, the first stimulus (condition 1) was presented 2 seconds before the first image was acquired, the second stimulus (condition 3) at the first image, and the third stimulus (fixation) at 4.5 seconds into the imaging run.
    -2.0 1
    0.0 3
    4.5 0

    For block design paradigms there will be a line for each image. The first column is the TR onset. The second column is the condition code for the current block. For example, if you have 2 12 second blocks and a TR of 3 seconds (starting with the baseline condition), the paradigm file will look like this:
    0 0
    3 0
    6 0
    9 0
    12 1
    15 1
    18 1
    21 1

    Such files can be created in Excel or Gnumeric (the latter is a spreadsheet program on Charlie, the Linux machine). You should have a parameter file for each run with your tutorial data because the tutorial data is event-related, which means that the timing will be unique for every run.

Make Analysis
  • Purpose: With this program ( you define how the raw or pre-processed data for each session will be averaged together. Data are not actually averaged at this stage. You may wish to analyze your study in more than one way. For example with and without motion correction, or you may wish to define your events based on subject response in one analysis and on stimulus type in another. Give each analysis a name ("analysisname") that will be used in subsequent processing steps. For multiple analyses, you will run mkanalysis-sess once for each analysis, giving each a different analysis name.

    This program creates a subdirectory called "analysisname" in the study directory and writes two files there: analysis.cfg (see further MGH documentation) and It also checks that all subject data is accessible and creates default scripts for the analysis.

  • Usage:
    Required Arguments
    -analysis analysisname : name of session-level functional analysis
    -TR TR : TR value in seconds
    -paradigm parname : name of paradigm file
    -designtype type : type of design. Legal values are: event-related,
    blocked, abblocked, retinotopy
    Optional Arguments
    -fsd FSD : functional subdirectory
    -funcstem stem : stem of functional data (default = f or fmc)
    -motioncor : use motion corrected data (see mc-sess)
    -runlistfile filename : file with list of runs to include in analysis
    -inorm : intensity normalize data (see inorm-sess)
    -fwhm FWHM : Full-Width/Half-Max for in-plane smoothing (mm)
    -nskip N : skip the first N time points in each run
    -tpef filename : time-point exclude file name
    -notrendfit : do not fit a linear trend
    Event-Related and Blocked Arguments
    -nconditions Nc : number of conditions (excluding fixation)
    -timewindow TW : event time window (sec)
    -prestim PS : event prestimulus window (sec)
    -TER TER : temporal estimation resolution (sec) (This makes a jagged HDR...perhaps it is best avoided)
    -gammafit gfDelta gfTau : assume IRF is a gamma function
    -taumax TauMax : maximum lag for noise autocor and whitening (sec)
    -timeoffset toffset : time offset (sec) to add to each paradigm file
    Retinotopy and AB Blocked Arguments (Required)
    -ncycles N : number of cycles in each run
    Other Optional Arguments
    -umask umask : set unix file permission mask
    -version : print version and exit
    -help : print help info

  • Command Examples (run from your parent subject directory):
    > -analysis video1 -TR 2.5 -paradigm par.txt -designtype event-related -funcstem fss -inorm -nskip 4 -nconditions 3 -timewindow 22.5 -prestim 5

    > -analysis video2 -TR 2.5 -paradigm par.txt -designtype event-related -funcstem fmc -inorm -fwhm 5 -nskip 4 -nconditions 3 -timewindow 22.5 -prestim 5

  • Explanation:
    • -analysis Here we create two new directories, video1 and video2 respectively, in the parent subject directory (e.g., /buddy/data/tutorial). For video1 we use the smoothed data, and for video2 we use the unsmoothed data.
    • -TR 2.5 sets the TR value in seconds.
    • -paradigm par.txt identifies the parameter file to use in making the analysis
    • -designtype event-related identifies the design type using one of the 4 legal values (i.e., "event-related").
    • -funcstem identifies the input volume functional stem (fss or f).
    • -inorm tells the program to use intensity normalized data.
    • -fwhm 5 has been added to the analysis of the unsmoothed data in video2 to add smoothing. Smoothing increases the signal to noise ratio...but if spatial-smoothing overdoes it...then mkanalysis is a better step to try it on. Spatial smoothing smooths within AND between planes but mkanalysis smooths only within plane.
    • -nskip 4 indicates that the first 4 timepoints should be disregarded (the first few scans are often not very good)
    • -nconditions 3 identifies the number of conditions (3), excluding fixation (i.e., the control)
    • -timewindow 22.5 This tells the program to display the graph of the HDR intensity line with an x-axis of 22.5 seconds.
    • -prestim 5 This tells the program to display the graph of the HDR intensity line beginning 5 seconds before the stimulus onset. The program will know the timing of the stimulus onset from the par file.
    • The prestim intensity values will eventually be used as baseline average values for further calculations. These intensity values will be subtracted from each intensity line so that all the lines can begin at the same level. tkmedit allows you to subtract prestim average (you'll see this graphed)
      Issue to learn about: BUT we don't yet know whether this subtracts the values from the text file output at some later step.
  • Output:
    • The command runs very quickly (~1 sec)
    • It generates a directory with the analysis name containing two small text files, and analysis.cfg that contain information about the analysis.

Select Average
  • Purpose: This program computes the average signal intensity maps for each condition for each individual subject. The output is stored in a subdirectory called "analysisname" under the bold subdirectory in the subject's directory. (The "analysisname" was defined in the mkanalysis-sess step, above). This average data can be further processed on an individual basis and/or can be used in group analysis.
  • Usage: selxavg-sess
    Required Arguments:
    -analysis analysisname : name of functional analysis
    Session Arguments (Required)
    -sf sessidfile ...
    -df srchdirfile ...
    -s sessid ...
    -d srchdir ...
    Session Arguments (Optional)
    -umask umask : set unix file permission mask
    -version : print version and exit

  • Command Examples:
    >selxavg-sess -analysis video1 -sf sessidfile -d .
    >selxavg-sess -analysis video2 -sf sessidfile -d .
  • Explanation:
    • -analysis identifies an analysisname for each directory where output will be stored: video1 and video2.
    • -sf sessidfile identifies the sessidfile used to find the subjects' directories.
    • -d /buddy/data/tutorial identifies the parent subject directory to work in.
  • Output:
    • Together, these two selxavg-sess commands took under 15 minutes to run. The video1 (fss) analysis took under 1 minute. The video2 (fmc) analysis took~13 minutes.
    • In each analyisdir (e.g., video1 and video2) selxavg creates the following files: allvres.mat, h.dat, omnibus.mat, and studydir.
    • selxavg-sess also creates an analysis directory in the subject's bold directory (e.g., for the tutorial, video1 and video2 directories are created in each bold directory). These secondary analysis directories contain the analysis.cfg and files (same as the copies in the main analysis dir), a full set (e.g., 25 pairs for our tutorial data) of h_ bfloat and hdr files, a full set of h-offset bfloat and hdr files, an h.bhdr and h.dat file (the h.dat file is the same as the one in the main analysis dir), h-offset.bhdr, h-sxa.log, X.mat and an omnibus directory.
    • The omnibus directory contains a full set of f_ bfloat and hdr files, a full set of fsig_ bfloat and hdr files, f.bhdr and fsig.bhdr.
    • selxavg creates a log file each time it run (e.g., selxavg-sess-bold-video1-0205111510.log, selxavg-sess-bold-video2-0205111512.log)

Make Contrast
  • Purpose: This program creates a file specifying which contrasts will be compared. The contrasts are not actually computed at this step. You will supply a name for this contrast file (e.g., "allvcon" below) which will be required at subsequent processing stages. This is analogous to the definition of contrasts in spm. For multiple contrasts, run mkcontrast-sess multiple times, defining a different contrast file name each time. The contrast is defined with respect to a particular analysis using the program options.
  • Usage: mkcontrast-sess
    -contrast contrastname : contrast name
    -analysis analysisname : name of session-level functional analysis
    -a a1 <-a a2> ... : positive contrast conditions
    -c c1 <-c c2> ... : negative contrast conditions
    -setwdelay : prompt for setting of delay weights
    -sumdelays : sum weighted delays
    -rmprestim : subtract prestimulus baseline
    -nosumconds : don't sum conditions (for f-test)
    -ircorr nircorr : correlate with gamma function
    -deltarange dmin dmax : range of delta for gamma function
    -taurange tmin tmax : range of tau for gamma function
    -umask umask : set unix file permission mask
    -scriptonly : don't run, just generate a script
    -version : print version and exit

  • Command Examples:
    >mkcontrast-sess -contrast allvcon -analysis video1 -a 1 -a 2 -a 3 -c 0 -ircorr 2 -deltarange 2 4
    >mkcontrast-sess -contrast allvcon -analysis video2 -a 1 -a 2 -a 3 -c 0 -ircorr 2 -deltarange 2 4
  • Explanation: Run mkcontrast-sess from your parent subject directory (e.g., /buddy/data/tutorial/).
    • -contrast allvcon creates a matlab *.mat file with the name you provide (e.g., allvcon.mat). The file holds contrast information, e.g., all conditions vs control in this case.
    • -analysis identifies the analysis directory where allvcon.mat will be stored.
    • -a 1 -a 2 -a 3 -c 0 weights each of the three conditions equally (1/3 each). and balances the summed conditions against the control weight (1).
    • -ircorr tells mkcontrast-sess to correlate hemodynamic responses with some number of ideal hemodynamic this case 2.
    • -deltarange flag specifies a 2 and 4 second offset from 0 for the two ideal hemodynamic responses. mkcontrast picks the best HDR offset and displays only that one. We don't know how to tell which one it has picked.
  • Output:
    • mkcontrast-sess runs very quickly (several seconds)
    • It creates a log file in the log directory: mkcontrast-sess.log
    • It creates a file (e.g. allvcon.mat) describing the contrasts in the named analysis directory (the analysis dir in the parent subject directory)

Statistics Grinder
  • Purpose: This program is used to create statistical, significance, and parametric maps for all of your subjects on an individual basis. The map is computed from a given analysis and contrast. Significance results are stored as log10 unless you specify otherwise using -pxform option.
  • Usage: stxgrinder-sess
    -analysis analysisname : name of session-level functional analysis
    -append code : sets analysis to analysisname-code
    -contrast contrastname : contrast name
    -pxform xform : <log10>, ln, none
    -space spacename : space in which to grind (native, tal, sph)
    -hemi hemisphere : with sph space <lh rh>
    -sf sessidfile ...
    -df srchdirfile ...
    -s sessid ...
    -d srchdir ...
    -scriptonly : don't run, just generate a script
    -version : print version and exit
    -umask umask : set unix file permission mask

  • Command Examples:
    >stxgrinder-sess -analysis video1 -contrast allvcon -sf sessidfile -d .
    >stxgrinder-sess -analysis video2 -contrast allvcon -sf sessidfile -d .
  • Explanation: stxgrinder-sess is run once for each of the two analyses. The contrast to be used is specified (you could have several contrasts). The program uses the sessidfile to find the subjects' directories. See previous commands for explanations of each flag.
  • Output:
    • stxgrinder-sess takes several minutes to run (~ 3 minutes for the tutorial data).
    • It creates a log file in the log directory (e.g., stxgrinder-sess-bold-video1-allvcon.log)
    • It creates a directory (allvcon) full of files in the secondary analysis directory (e.g., video1 in each subject's bold directory). This contains several full sets of bfloat and hdr files: f_, fsig_, iminsig_, minsig_, sig_, and t_. Each set has an accompanying *.bhdr file. allvcon.mat (this is the same as allvcon.mat in the main analysis directory) and sig_stxg.log are also in this directory.

Register Functionals to Structurals
  • Purpose: This step aligns the functional images to the 3D structural. In this case, we have not talairached the data.
  • Usage: Tkmedit Online Guide
  • Command Example :
    >tkmedit e25996_anat orig -overlay /buddy/data/tutorial/e25996/bold/001/fmc -overlay-reg /buddy/data/tutorial/e25996/bold/register.dat -register
  • Explanation: This command brings up the tkmedit gui interface with your 3dvolume in e26154_anat orig (the COR file location).
    • -overlay brings up the functional data from the specified directory. Finally, the
    • -overlay-reg flag is followed by the path to the register.dat file (currently the unity matrix) the
    • -register flag enables functional registration.

GUI Interface

Choose View-->Configure-->Functional overlay (set thresholds to minimum=600, midpoint=700, slope=.001, although these #'s can vary. The goal is to control the size and shape of the displayed activation so that it looks like the picture to the left, even if not yet aligned to the anatomical image). The images will be rotated 90 degrees out of alignment with each other (this is true at least for the tutorial data), you must rotate and align them:
Tools-->fMRI-->register functional overlay. Click the sagittal orientation button (on the main gui menu pictured above), ...rotate: 90 degrees (then click the right arrow on the rotate menu). Next do finer grained alignment: translate (in mm...inplane sliding) until they appear to be nicely aligned.
When you are satisfied with the registration, choose: Save-->overlay registration. This alters the register.dat file,and saves the original as register.dat.1

Here you see the functional data overlayed on the 3D structural and aligned.

    • Make sure register.dat file is no longer the unity matrix to confirm that your changes have been made.
    • This does the registration for all runs for one subject. Repeat Functional-Structural registration for each subject.

Check your Work with tkmedit-sess
  • Purpose: To look at significance maps on non-talairached 3ds.
  • Usage: tkmedit-sess
    -volid FSVolId : name of volume under subject/mri
    -analysis analysisname : name of session-level functional analysis
    -s sessid (only one allowed)
    -isxavg method (fixed or random)
    -contrast contrastname : contrast name
    -map mapname : <sig>, minsig, iminsig, t
    -mapanalysis analysisname : name of session-level functional analysis
    -mapsess sessid from which to get the map
    -mapisxavg method
    -space spacename : <native> or tal
    -d sessdir ...
    -sessanat : use session 3d anatomical instead of recon
    -fsd dir : functional subdirectory (bold)
    -asd dir : anatomical subdirectory (3danat)
    -scriptonly : don't run, just generate a script
    -umask umask : set unix file permission mask
    -version : print version and exit
  • Command Example :
    >tkmedit-sess -analysis video1 -contrast allvcon -map minsig -s ./e26154 -d .
    One can look at the "green squares" of the posclust output by loading labels (e.g., /e26154_anat/label/posclust-0001.label; this option is available in the gui interface). In addition, time graphing options are available in the gui interface.
    Further options to learn about: Look into loading a label, then graphing current selection, then printing timecourse summary--this can potentially give you averaged, baseline subtracted data for all voxels in a cluster eliminating a great deal of excel editing.
    NOTE--you can also create your own cluster by painting it--can select width of brush& 3d as well-- Tools--fmri--graph current selection, tools--fmri--print timecourse summary to file, and should save the label too)!
  • Explanation: All runs for an individual subject are now averaged together in this view of the data. Green square mark the location of each cluster.
    • -analysis video1 indicates which analysis directory to use
    • -contrast allvcon indicates which contrast file to use in the analysis directory "video2"
    • -space tal this dir contains the results of the post-talairaching statsgrinder run, specifically "maps" like minsig of the talairached data. The maps are then used in this step.
    • -map minsig
    • -s ./e26154
    • -d /buddy/data/tutorial this identifies the parent subject directory

MRI Volume Clustering
  • Purpose: This command identifies clusters of activations that meet specified criteria. It then creates images, summary files and even numbered labels for the output clusters.
  • Usage: mri_volcluster
    --in input volid
    --in_type file format
    --frame frameno <0>
    --reg register.dat
    --thmin minthresh
    --thmax maxthresh (default is infinity)
    --sign <abs>, neg, pos
    --minsize minimum volume (mm^3)
    --minsizevox minimum volume (voxels)
    --mindist distance threshold <0>
    --allowdiag : define contiguity to include diagonal
    --mask mask volid (same dim as input)
    --mask_type file format
    --maskframe frameno <0>
    --maskthresh upper threshold
    --masksign <abs>, neg, pos
    --outmask final binary mask
    --outmask_type file format
    --sum file : text summary file

    --out outupt volid
    --out_type file format
    --ocn output cluster number volid
    --ocn_type file format
    --label label file
    --nlabelcluster n : save nth cluster in label
    --labelbase base : save all clusters under base-NNNN.label

    --synth synthfunc (uniform,loguniform,gaussian)
    --help : how to use this program

  • Command Example:
    > mri_volcluster --in /buddy/data/tutorial/e25996/bold/video2/tal/allvcon/minsig --in_type bfloat --reg /buddy/data/tutorial/e25996/bold/video2/tal/register.dat --thmin 9 --sign pos --minsize 640 --allowdiag --out /buddy/data/tutorial/e25996/bold/video2/tal/allvcon/mpclus_minsig --out_type bfloat --ocn /buddy/data/tutorial/e25996/bold/video2/tal/allvcon/mpclusnum_minsig --ocn_type bfloat --labelbase /buddy/data/tutorial/e25996/bold/video2/tal/allvcon/me25996 --sum /buddy/data/tutorial/e25996/bold/video2/tal/allvcon/me25996sum
  • Explanation:
    • By this point we have rejected the analysis in video1 as having too much smoothing. So we only analyze video2.
    • --in /buddy/data/tutorial/e25996/bold/video2/tal/allvcon/minsig specifies that minsig, one of several statistical maps available for clustering, should be used.
    • --in_type bfloat specifies that the type of map file being used is the bfloat (the maps are created by stxgrinder)
    • --reg /buddy/data/tutorial/e25996/bold/video2/tal/register.dat identifies the registration.dat file that should be used for the procedure.
    • --thmin 9 sets the p-value to .000000001 for clustering. The value should represent the absolute log10 of the p-value. Thus:
      -thmin 1.3010 =.05,
      -thmin 2 =.01,
      -thmin 3=.001,
      -thmin 4 =.0001
    • --sign pos tells the program to represent only those clusters that are positively correlated to the experimental paradigm and not those that are negatively correlated.
    • --minsize 640 sets up a cluster of 10 voxels. [Although we often use the equivalent of 3 voxels in our clustering, (voxels were resampled to 4x4x4 [64 mm cubed], so 3x64=192 mm cubed), it is probably better to use 640 (10 voxels) for this tutorial data, because 192 mm-cubed produces an awful lot of clusters.]
    • --allowdiag allows contiguity between voxels to be defined as including voxels that touch on corners only.
    • --out will put a map of activations that meet the criteria into the specified file, /buddy/data/tutorial/e25996/bold/video2/tal/allvcon/mpclus_minsig (name it whatever you want...will get one per slice [e.g., 25) why is this necessary if we have ocn?).
    • --out_type bfloat tells the program that the output images should be bfloats.
    • --ocn will number each cluster output cluster number map in .bfloat format (one for each slice, e.g., 25, but labelled)
    • --ocn_type will be bfloat.
    • --labelbase followed by prefix to add to each ocn label for the clusters (one file for each cluster; later these label files can be loaded to view each green square that demarcates an ROI)
    • --sum is the single summary file me25996sum which will hold summary info on each cluster by its ocn number.
  • Run mri_volcluster once for each subject

Cluster by Cluster Labeling
  • Purpose: This command computes the average HDR for a given cluster. These averages are useful, because before running this for a cluster, you see a different HDR in the timecourse graph for each different voxel. The file h.txt contains the average for an ROI defined by your cluster. Run this once for each labeled cluster that you want to see information about. can look at output from above (where?)--posclusnum in tkmedit--in association with the summary file above (me25996sum) to locate ROIs. ....Maybe w good cluster labeling, this could be run as a session (say supFrntGyrus on every subject who had the same label??
  • Usage: func2roi-sess
    -roidef name : name of ROI definition
    -analysis name : source is averaged data from analysis
    -raw : source is raw data
    -rawfsd : functional subdirectory for raw data (bold)
    -motioncor : use motion corrected raw data (with -raw)
    -rawstem : stem of raw data (f or fmc)
    -rawrlf : run list file for raw data
    -noraw : don't process raw data
    -sesslabel name : use name.label found in sessid/labels
    -anatlabel name : use name.label found in SUBJECTS_DIR/subject/label
    -labelfile file : give full path of label file
    -labelspace space
    -labelspacexfm xfm file found in mri/transforms (talairach.xfm)
    -maskcontrast contrast: contrast to use for mask
    -maskthresh threshold
    -masktail tail : thresholding tail (<abs>, pos, neg)
    -maskmap map : map to use for mask <sig>
    -maskframe frame : 0-based frame number in map <0>
    -masksessid sessid : sessid of mask (default is that of source)
    -maskanalysis name : analysis of mask (default is -analysis)
    -maskspace space : space of mask (<native> or tal)
    -maskisxavg effect: fixed or random (when masksessid is a group)
    -float2int method: method = <tkreg>,round,floor
    -sf sessidfile ...
    -df srchdirfile ...
    -s sessid ...
    -d srchdir ...
    -umask umask : set unix file permission mask
    -scriptonly : don't run, just generate a script
    -version : print version and exit

  • Command Example:
    >func2roi-sess -analysis video2 -roidef minsig_posclust -noraw -labelfile
    /buddy/data/tutorial/practiceSubject/posclust-0001.label -s e26154 -d .

  • Explanation: the output of this command is saved to bold/video2/minsig_posclust/h-offset as bvolume...once for each subject label of interest...maybe 1/2 a dozen??

  • Purpose: To talairach anatomical data.
  • Command Example:
    >cp e26154_anat/mri/orig/* e26154_anat/mri/T1/
    >talairach e26154_anat
  • Explanation:
    You'll need to copy *.COR files from the orig directory to the T1 directory
    in order to talairach them.
    The talairach command outputs talairach.xfm in the /mri/transforms subdirectory. (e.g., /buddy/data/tutorial/e26154_anat/mri/transforms/talairach.xfm).

Talairaching Functionals
  • Purpose: To talairach functional data (uses register.dat file and /mri/transforms/talairach.xfm files)
  • Usage: func2tal-sess
    Required Arguments:
    -res mm : talairach resolution in mm (1,2,4,8)
    -analysis analysisname : name of session-level functional analysis
    Optional Arguments:
    -spacedir dir : default is tal
    -xfm xfmfile : xfm file relative to subjid/mri/transforms
    -umask umask : set unix file permission mask
    -version : print version and exit
    Session Arguments (Required):
    -sf sessidfile
    -df srchdirfile
    -s sessid
    -d srchdir
  • Command Example:
    >func2tal-sess -res 4 -analysis video2 -sf sessidfile -d .
  • Explanation:
    -res 4 talairach resolution set to 4 mm. -analysis video2 name of analysis directory. -sf uses the sessionidfile to identify which subjects to run the process on. However, one could use -s and specify an individual subject instead.

Stats Grinder

Purpose and Usage

Run stxgrinder on talairached selxavg output
>stxgrinder-sess -analysis video2 -contrast allvcon -space tal -s ./e26154 -d

Check your Work with tkmedit-sess

Purpose: Bring up talaraiched functional data on talaraich (MNI) brain:
Command Example:
>tkmedit-sess -analysis video2 -contrast allvcon -space tal -map minsig -s ./e26154 -d .

-space tal
this dir contains the results of the post-talairaching statsgrinder run, specifically "maps" like minsig of the talairached data. The maps are then used in this step.

Fixed-effect group analysis
  • Purpose: The fixed-effect group analysis is a "quick" test that treats a group of subjects as one subject. The criteria are thus not as stringent as for the random effects group analysis (which treats each subject separately).
  • Usage: isxavg-fe-sess
    -analysis analysisname : session-level functional analysis name
    -group groupname : name of group
    -space spacename : space in which to average (native, tal, sph)
    -hemi hemisphere : with sph space <lh rh>
    -trunc sign : truncation (pos or neg; neg = set neg vals to 0)
    -sf sessidfile ...
    -df srchdirfile ...
    -s sessid ...
    -d srchdir ...
    -umask umask : set unix file permission mask
    -scriptonly : don't run, just generate a script
    -version : print version and exit
  • Command Example 1:
    >isxavg-fe-sess -analysis video2 -group fixedgroup -space tal -trunc neg -sf sessidfile -d .
  • Explanation: -analysis video2 identifies the analysis directory to use (does it get files from here or put files in here? -group fixedgroup names a directory to be created at the top level to contain the fixed-effect group analyses -space tal
    -trunc neg
    this option ignores/truncates all negative correlations, i.e., it shows the positive correlations only in the output.
    -sf sessidfile
    identifies the sessionidfile that the command uses to find all the subjects.
    -d /buddy/data/tutorial
    identifies the parent subject directory to work in. Why no -contrast allvcon here (as compared to the random effects analysis)?

    >stxgrinder-sess -analysis video2 -s fixedgroup -space tal -contrast allvcon -sf sessidfile -d .
  • Explanation:
  • Command: Vol clustering with fixed-effect group analysis output
  • >mri_volcluster --in /buddy/data/tutorial/fixedgroup/bold/video25/tal-ffx/allvcon/minsig --in_type bfloat --reg /buddy/data/tutorial/fixedgroup/bold/video2/tal-ffx/register.dat --thmin 4 --sign pos --minsize 640 --allowdiag --out /buddy/data/tutorial/fixedgroup/bold/video2/tal-ffx/allvcon/4pclus_minsig --out_type bfloat --labelbase
    /buddy/data/tutorial/fixedgroup/bold/video2/tal-ffx/allvcon/fixed --sum

  • Displaying the Positive Cluster Map from fixed group analysis
    >tkmedit-sess -contrast allvcon -analysis video2 -map 4pclus_minsig -space tal -isxavg fixed -s fixedgroup -d .
  • Func2ROI with Fixed Group Analysis
    1) copy all the h and h offset files from /buddy/data/tutorial/fixedgroup/bold/video2/tal-ffx to /buddy/data/tutorial/fixedgroup/bold/video2/.

    2) You then need to create an file and place it in this directory also. The file should read something like this (although of course adapt it to your case. Also note that if you're running the same analysis for the fixed effect group as for what?...I'm missing something here. you can just cp file from a subject does not exist until you create it here?):
  • analysis video2
    TR 2.5
    designtype event-related
    nconditions 3
    parname par.txt
    fsd bold
    funcstem fmc
  • 3) cp register.dat file that was in here:/buddy/data/tutorial/fixedgroup/bold/video2/tal-ffx to fixedgroup/bold.
  • 4) Make a subjectname file in fixedgroup directory with "talairach" in the body (no "s).
  • 5) Use func2roi-sess command:
  • >func2roi-sess -analysis video5 -labelspace tal -roidef ./tal-ffx/allvcon/roi_0002 -noraw -labelfile /buddy/data/tutorial/fixedgroup/bold/video2/tal-ffx/allvcon/fixed-0002.label -s fixedgroup -d .
  • Makes fixedgroup/bold/video2/tal-ffx/allvcon/roi_0002 directory and creates a set of mask files and the h.txt file...
  • TO VIEW:
    Regular minsig map:
    >tkmedit-sess -contrast allvcon -analysis video5 -map minsig -space tal -isxavg fixed -s fixedgroup -d .
  • TO view label:
    >tkmedit-sess -contrast allvcon -analysis video5 -map roi_0002/mask -space tal -isxavg fixed -s fixedgroup -d .

Random Effects Group Analysis
  • Purpose:
  • Usage: isxavg-re-sess
    -analysis analysisname : session-level functional analysis name
    -group groupname : name of group
    -space spacename : space in which to average (native, tal, sph)
    -hemi hemisphere : with sph space <lh rh>
    -contrast contrastname: contrast name
    -pctsigch : use percent signal change
    -nojackknife : do not use jackknifing
    -trunc sign : truncation (pos or neg; neg = set neg vals to 0)
    -sf sessidfile ...
    -df srchdirfile ...
    -s sessid ...
    -d srchdir ...
    -scriptonly : don't run, just generate a script
    -version : print version and exit
  • Command:

    >isxavg-re-sess -analysis video2 -contrast allvcon -group randomgroup -space tal -sf sessidfile -d .
  • Explanation:
  • -analysis video2 identifies the analysis directory to use
  • -contrast allvcon identifies the contrast to be used .
  • -group randomgroup names a directory to be created at the top level to contain the random-effects group analyses
  • -space tal
  • -trunc neg ??
  • -sf sessidfile identifies the sessionidfile that the command uses to find all the subjects.
  • -d . identifies the parent subject directory to work in.

Compare Results
  • Purpose: Compare the results of the random-group and fixed-group analyses.
  • Commands:
    >tkmedit-sess -analysis video2 -contrast allvcon -space tal -map msig -isxavg random -s randomgroup -d .
    >tkmedit-sess -analysis video2 -contrast allvcon -space tal -map pclus_minsig -isxavg fixed -s fixedgroup -d .
  • Explanation:

View Random Effects Clusters with Fixed Effects Analysis Headers
  • Purpose:
  • Commands:
    >tkmedit-sess -analysis video2 -contrast allvcon -space tal -map minsig -isxavg random -s /buddy/data/tutorial/randomgroup -d .
  • GUI:
    File->Load Timecourse: /buddy/data/tutorial/fixedgroup/bold/video2/tal-ffx/allvcon
    stem: minsig
    Register: /buddy/data/tutorial/fixedgroup/bold/video2/tal-ffx/allvcon/register.dat
  • Explanation: