Print

Print


i tried the LOO option for manual training and i get the following error , did you get the same ?

Error using call_fsl (line 36)
FSL call (/bin/sh -c '. /usr/share/fsl/5.0/etc/
fslconf/fsl.sh; FSLOUTPUTTYPE=NIFTI_PAIR; export FSLOUTPUTTYPE; $FSLDIR/bin/fslmaths 6/..feat/fix/hr2exf /tmp/tp20010898702001') failed, Cannot open volume 6/..feat/fix/hr2exf for reading!

On Tue, Nov 11, 2014 at 4:04 AM, David V. Smith <[log in to unmask]> wrote:
I suspect FAST is being used during the feature generation process. You should check out the articles describing FIX for additional details:

G. Salimi-Khorshidi, G. Douaud, C.F. Beckmann, M.F. Glasser, L. Griffanti S.M. Smith. Automatic denoising of functional MRI data: Combining independent component analysis and hierarchical fusion of classifiers. NeuroImage, 90:449-68, 2014

L. Griffanti, G. Salimi-Khorshidi, C.F. Beckmann, E.J. Auerbach, G. Douaud, C.E. Sexton, E. Zsoldos, K. Ebmeier, N. Filippini, C.E. Mackay, S. Moeller, J.G. Xu, E. Yacoub, G. Baselli, K. Ugurbil, K.L. Miller, and S.M. Smith. ICA-based artefact removal and accelerated fMRI acquisition for improved resting state network imaging. NeuroImage, 95:232-47, 2014

Does the LOO option (-l) work for you? I still can’t figure out the error I’m having with it (though I can now generate the *.Rdata file after fixing the issue Steve identified):

leaving-out: /mnt/delgadolab/disk02/labdata02/AffInf01/FSL/203/run1.feat
/bin/sh: 1: 302:: not found

Thanks,
David


On Nov 10, 2014, at 5:11 PM, Alireza Sedghi <[log in to unmask]> wrote:

Thanks steve , it finally worked ( my fault in setting the MATLAB_ROOT)

i just wonder why it performs FAST ?
do you have any idea why it does segmentation ?

On Tue, Nov 11, 2014 at 12:13 AM, David V. Smith <[log in to unmask]> wrote:
Hi,

I suspect one or both of the variables below are not defined correctly in your settings.sh file:

FSL_FIX_MATLAB_ROOT
FSL_FIX_MATLAB

That should fix the path error in your output ("/usr/local/MATLAB/2014a/bin/matlab not found!”).

CHeers,
David



On Nov 10, 2014, at 3:24 PM, Alireza Sedghi <[log in to unmask]> wrote:

Hi David thanks for your reply

i get the following error when applying fix on the full first level analysis

~/fix1.06/fix ..feat ~/fix1.06/training_files/Standard.RData 20
FIX Feature extraction for Melodic output directory: ..feat
 create edge masks
 run FAST
 registration of standard space masks
 extract features
/usr/local/MATLAB/2014a/bin/matlab not found!
/home/fix1.06/call_matlab.sh [-z <mode>] [-o <options>] [-p <path>] [-h <fix path>]
    [-c <mcr path>] [-b <binaries path>] [-l <logfile>]
    {-r <"MATLAB code">|-m <script>|-f <function> [arguments...]}

    -z <mode>:    This overrides the environment variable
        FSL_FIX_MATLAB_MODE which would    otherwise control this.
            0 = Use compiled MATLAB function specified by
                -m <function name>
            1 = Use MATLAB - requires -r or -m option
            2 = Use Octave - requires -r or -m option
    -r <"MATLAB code">: Evaluate the double-quoted MATLAB/Octave
        code. Equivalent to 'matlab -r "MATLAB code"' or
        'octave --eval "MATLAB code"'
        Ensure that any double quotes in this string are escaped
        with '\'.
    -m <script>:
        -z = 0 or not provided, run as -z = 1
        -z = 1|2 - Run <script> in MATLAB/Octave
            Equivalent to 'matlab \< <script>.m' or 'octave <script>.m'
    -f <function> [arguments]:
        -z = 0 - Call compiled MATLAB <function>
            Arguments to be passed are given in the remainder of the
            command line.
        -z = 1|2 - Run <function> script in MATLAB/Octave
            Equivalent to 'matlab -r "<function>([arguments])"' or
            'octave --eval "<function>([arguments])"'
            Arguments to be passed are given in the remainder of the
            command line.
    -o <options>: Double quoted list of MATLAB/Octave command line
        options. Defaults to ${FSL_FIX_MLOPTS} or ${FSL_FIX_OCOPTS}
        respectively
    -p <path>: Path to MATLAB/Octave startup script/binary.
        Defaults to: ${FSL_FIX_MATLAB} or ${FSL_FIX_OCTAVE}
        respectively
    -h <path>: Path to folder containing the Fix programs
    -w <path>: Path to FSL MATLAB helper functions, defaults to
            /usr/share/fsl/5.0/etc/matlab
    -c <mcr path>: Path to MATLAB Compiler Runtime for the version
        of MATLAB Fix was compiled under. Defaults to ${FSL_FIX_MCRROOT}
    -b <compiled path>: Path to the folder containing the compiled MATLAB
        functions. Defaults to ${FSL_FIX_MLCDIR}
    -l <logfile>: Append console and error output to <logfile>\n

FIX Classifying components in Melodic directory: ..feat using training file: /home/fix1.06/training_files/Standard.RData and threshold 20
FIX features not already generated, so creating them now.
FIX Feature extraction for Melodic output directory: ..feat
 create edge masks
 run FAST


actually i can see the matlab file in the /usr/local/MATLAB/2014a/bin/

it seems that it has entered an infinite loop ! it wants to classify , it does not find features , wants to create , there is no matlab , wants to classify ......


On Mon, Nov 10, 2014 at 7:25 PM, David V. Smith <[log in to unmask]> wrote:
Hi,

Are you getting any errors? I might be misunderstanding something, but it sounds like you don’t really need to use the compiler or change anything beyond the instructions in the README file (e.g., the R libraries and the paths). I’m not using the compiler, and FIX generally works for me me, aside from the error I mentioned earlier.

Aside from the defaults, the only thing I’ve done is hardcode the FSL_FIXDIR in my ~./bashrc file since the settings.sh file was not populating that variable correctly for me.

FSL_FIXDIR=/usr/local/fix/fix1.06
PATH=$PATH:$FSL_FIXDIR
export PATH FSL_FIXDIR

Again, pretty much everything works for me except the LOO accuracy estimation, so I’m curious to hear if you’re having the same error or something different.

Cheers,
David


On Nov 10, 2014, at 9:40 AM, Alireza Sedghi <[log in to unmask]> wrote:

thanks
do i need to point to matlab compiler runtime if i have matlab program installed ?

about the settings related to matlab startup function

Set this to the MATLAB start-up options. Typically you will
# want to disable Java, display output, the desktop environment
# and the splash screen
FSL_FIX_MLOPTS="-nojvm -nodisplay -nodesktop -nosplash"

# Set this to the MATLAB 'evaluate string' option
FSL_FIX_MLEVAL="-r"
# Set this to the pass in file option
FSL_FIX_MLFILE="\<"

what should i do ?

On Mon, Nov 10, 2014 at 4:28 PM, Stephen Smith <[log in to unmask]> wrote:
Hi - if you're only using NIFTI data (and not CIFTI data from the HCP) then you don't need to worry about those settings.
Cheers.


On 10 Nov 2014, at 12:55, Alireza Sedghi <[log in to unmask]> wrote:


On Sun, Nov 9, 2014 at 10:26 PM, David V. Smith <[log in to unmask]> wrote:
describing

thanks for your response
actualy i have read readme and i have matlab with signal toolbox installed ( so i do not need to isntalle MCRInstaller ? )

 i cannot understand this part in the Readme:

Setup for using inside HCP pipelines (only needed for HCP):
Ensure the FSL_FIX_CIFTIRW and FSL_FIX_WBC variables in settings.sh are pointed
at your HCP Workbench MATLAB CIFTIRW and Workbench folders respectively.





---------------------------------------------------------------------------
Stephen M. Smith, Professor of Biomedical Engineering
Associate Director,  Oxford University FMRIB Centre

FMRIB, JR Hospital, Headington, Oxford  OX3 9DU, UK
+44 (0) 1865 222726  (fax 222717)
[log in to unmask]    http://www.fmrib.ox.ac.uk/~steve
---------------------------------------------------------------------------

Stop the cultural destruction of Tibet







--

Sincerely,



Alireza Sedghi

Graduate Research Assistant 
Laboratory of Machine Vision and Biomedical Image Processing
Department of Biomedical Engineering
K.N.Toosi University of Technology


----------------------------------------------
David V. Smith, Ph.D.
Postdoctoral Fellow, Delgado Lab
Department of Psychology
Rutgers University
Newark, NJ 07102
----------------------------------------------




--

Sincerely,



Alireza Sedghi

Graduate Research Assistant 
Laboratory of Machine Vision and Biomedical Image Processing
Department of Biomedical Engineering
K.N.Toosi University of Technology





--

Sincerely,



Alireza Sedghi

Graduate Research Assistant 
Laboratory of Machine Vision and Biomedical Image Processing
Department of Biomedical Engineering
K.N.Toosi University of Technology