[MINC-users] using EMMA: possible without Mex?

[Andrew Janke]

On Wed, Apr 9, 2008 at 12:00 AM, Audette, Michel
<Michel.Audette at medizin.uni-leipzig.de> wrote:
>  What I want is the voxel values at specific tag points, once Matlab reads the tag files, so that I can visualize these tag points in feature space.

Hi Michel,

I once wrote a program called mincsample for doing this sort of thing.
 The only spanner in the works is that I primarily wrote it for doing
random samples from files based upon a mask.  It allows you to do
things like this:

gordon:~$ mincsample -random_samples 10 -coords ~/data/me/a.mnc
-97.553589999999999804  -10.734777364705921698  -108.63487000000000648
 1.9928282597085527073
-97.553589999999999804  107.16522058039204524   -88.947370000000006485
 1.0070954451819638109
-81.616089999999999804  117.06522040784302874   -75.822370000000006485
 7.0038910505836575737
-1.9285899999999998045  62.165221364705800511   -66.447370000000006485
 46.982528419928279106
-23.491089999999999804  131.46522015686261398   -66.447370000000006485
 2.0080872816052490748
-31.928589999999999804  67.565221270588153857   -9.2598700000000064847
 38.989852750438693363
97.446410000000000196   90.965220862744999408   4.8026299999999935153
 11.996643015182726799
-111.6160899999999998   84.665220972548922873   74.177629999999993515
 1.9958800640878919808
-115.3660899999999998   58.565221427450900649   106.99012999999999352
 1.0070954451819638109
36.508910000000000196   -28.734777050980419233  122.92762999999999352   0

Where here the output is the coordinate followed by the value.

gordon:~$ mincsample -help
Command-specific options:
General options:
 -version:        print version info and exit.
 -verbose:        print out extra information.
 -quiet:          be very quiet.
 -clobber:        clobber existing files.
 -max_buffer:     maximum size of buffers (in kbytes)
                Default value: 4096
 -mask:           select voxels within the specified mask
 -mask_val:       mask value to use
                Default value: 1

Sampling Types:
 -all:            sample all the data (Default)
 -random_seed:    Random seed to use (use to get reproducible runs)
Default: use tv_usec
                Default value: -1
 -random_samples: take # random samples from the input data
                Default value: 0

Output Options:
 -sample:         Output a <mask.mnc> file of chosen points
 -outfile:        <file> for output data (Default: stdout)
 -append:         append output data to existing file
 -ascii:          Write out data as ascii strings (default)
 -double:         Write out data as double precision floating-point values
 -coords:         Write out world co-ordinates as well as values

Generic options for all commands:
 -help:           Print summary of command-line options and abort
 -version:        Print version number of program and exit

Usage: mincsample [options] <in1.mnc> <in2.mnc> ...
       mincsample -help

The help output is above, so in essence if you can convert your tag
points to a mask then all you would need to do is this:

   mincsample -coords -ascii -mask tag_mask.mnc infile.mnc

if you wanted output like the above example. If you wanted to save a
bit of space and output in double then just swap the -ascii flag with
a -double.   You can find mincsample (in deb and source form) on
packages.bic.mni.mcgill.ca.

Let me know if you need any more help with it, the simple answer here
is that I should add tag support to mincsample (Anders Rodell is
already requesting this!) but just haven't got to it yet... :)


a