convert_hvolume

 convert_hvolume -- convert the "h_*" files to a useful format

 Usage:  [hMeans, hStds, numSamples] = convert_hvolume(hvol, numCond) 

 As a result of the "selxavg" command in FS-FAST, "h_*" and 
 "h-offset_*" volumes are created containing the mean deviation from
 baseline for each condition at each voxel and the offset value at each voxel, 
 respectively.  In what I presume was a way to use the "bfloat" format
 to store these results, the values at each voxel in the "h_*" files
 are concatinated together, with standard deviations following means, and
 each condition following the previous one.  Thus, the dimensions of the
 hvol variable are [numSlices, numRows, numCols, 2*numCond*numSamples],
 where numSamples is how many timepoints each condition was sampled.
 convert_hvolume will take an hvol as input and split out the mean and
 standard deviation values into hMeans and hStds, both of dimension
 [numSlices, numRows, numCols, numConds, numSamples].

 This function was originally written by Doug Greve and can be found in the
 Freesurfer distribution in fsfast/toolbox/yak.m; it has been modified
 slightly from its original form.

 $Id: convert_hvolume.m,v 1.4 2003/09/15 15:52:18 nknouf Exp $

0001 function [hMeans, hStds, numSamples] = convert_hvolume(hvol, numCond)
0002 % convert_hvolume -- convert the "h_*" files to a useful format
0003 %
0004 % Usage:  [hMeans, hStds, numSamples] = convert_hvolume(hvol, numCond)
0005 %
0006 % As a result of the "selxavg" command in FS-FAST, "h_*" and
0007 % "h-offset_*" volumes are created containing the mean deviation from
0008 % baseline for each condition at each voxel and the offset value at each voxel,
0009 % respectively.  In what I presume was a way to use the "bfloat" format
0010 % to store these results, the values at each voxel in the "h_*" files
0011 % are concatinated together, with standard deviations following means, and
0012 % each condition following the previous one.  Thus, the dimensions of the
0013 % hvol variable are [numSlices, numRows, numCols, 2*numCond*numSamples],
0014 % where numSamples is how many timepoints each condition was sampled.
0015 % convert_hvolume will take an hvol as input and split out the mean and
0016 % standard deviation values into hMeans and hStds, both of dimension
0017 % [numSlices, numRows, numCols, numConds, numSamples].
0018 %
0019 % This function was originally written by Doug Greve and can be found in the
0020 % Freesurfer distribution in fsfast/toolbox/yak.m; it has been modified
0021 % slightly from its original form.
0022 %
0023 % $Id: convert_hvolume.m,v 1.4 2003/09/15 15:52:18 nknouf Exp $
0024 
0025     %% add one to include fixation
0026         numCond = numCond + 1;
0027 
0028         [numSlices, numRows, numCols, numTimepoints] = size(hvol);
0029 
0030         %% find out the number of samples by taking the
0031     %% total number of timepoints
0032         %% in the h volume and dividing by 2*numCond (since the hvol
0033         %% holds both the mean and the standard deviation)
0034         numSamples = numTimepoints/(2*numCond);
0035 
0036         %% reshaping and permutations to get the conditions in the
0037         %% correct order
0038     %% we clear the temporary hvol* variables after we are through,
0039     %% as we can run into matlab "out of memory" errors on large
0040     %% or hi-res datasets
0041         hvol2 = permute(hvol, [4 3 2 1]);
0042         clear hvol;
0043         hvol3 = reshape(hvol2, [numSamples, 2, numCond, numCols, numRows, numSlices]);
0044     clear hvol2;
0045 
0046         hvol4 = permute(hvol3, [6 5 4 3 1 2]);
0047         clear hvol3;
0048 
0049         %% store the means and standard deviations separately
0050         hMeans = (hvol4(:,:,:,:,:,1));
0051         hStds = (hvol4(:,:,:,:,:,2));
0052 return;