Program name: vancova

Function: Voxelwise analysis of covariance on image data

Author: Babak A. Ardekani, Ph.D.

 

Usage:

            vancova [–v/–verbose] [–dataMask dataMaskCode] [–FWMH  FWHM]

[–o/–output prefix] [–m/–mask mask] [–x  column] [–y  row] [–z  slice] 

[–nozero] –d/–data dataFile  –dataType dataTypeCode  –c/–contrast contrastFile

 

Required arguments:

 

–d or –data  dataFile 

Specifies a text file (dataFile) containing a data table used in the ANCOVA analyses.  Columns of this table can be either numerical or list image files.  However, the first column is reserved for the dependent variable, and must contain images.  All images are expected to be in ANALYZE format and of type ‘short’.  An example dataFile may be as follows:

 

FA/C375FA.hdr      1.0      1      0      1.0      WM/C375WM.hdr

FA/C380FA.hdr      2.0      1      0      3.0      WM/C380WM.hdr

FA/C499FA.hdr      3.0      0      1      2.0      WM/C499WM.hdr

FA/C510FA.hdr      4.0      0      1      6.0      WM/C510WM.hdr

FA/C520FA.hdr      1.0      1      0      5.0      WM/C520WM.hdr

FA/C557FA.hdr      2.0      1      0      8.0      WM/C557WM.hdr

FA/C661FA.hdr      3.0      0      1      7.0      WM/C661WM.hdr

FA/C663FA.hdr      5.0      0      1      9.0      WM/C663WM.hdr

FA/C671FA.hdr      3.0      1      0      9.0      WM/C671WM.hdr

FA/C683FA.hdr      1.0      1      0      1.0      WM/C683WM.hdr

FA/C704FA.hdr      1.0      1      0      3.0      WM/C704WM.hdr

FA/C713FA.hdr      1.0      1      0      3.0      WM/C713WM.hdr

FA/C741FA.hdr      1.0      0      1      8.0      WM/C741WM.hdr

FA/C745FA.hdr      1.0      0      1      7.0      WM/C745WM.hdr

 

This files has 6 columns. Columns 1 and 6 contain image files. Columns a 2, 3, 4, and 5 are numerical.  The first column lists 14 images ‘C*FA.hdr’ in a directory named ‘FA’.  These images contain the dependent variable for each ANCOVA analysis.  The 6th columns lists 14 images ‘C*WM.hdr’ in a directory named ‘WM”.  These images contain one of the independent variables at each voxel.  All images are assumed to be spatially registered, and have equal matrix size and number of slices.  The first column specifies the dependent variable.  The subsequent columns contain either effects of interest or covariates.  The program runs a voxelwise ANCOVA, where at each voxel the image columns (columns 1 and 6 in this example) are replaced by the image values at the voxel being analyzed.  The advantage of this program over similar programs (e.g., SPM) is that it allows the linear model to change from one voxel to another.

 

–dataType  dataTypeCode

dataTypeCode is a string exclusively of characters ‘i’ or ‘n’ that specifies the type of data (numerical or image list) contained in the datafile, where ‘i’ signifies image list columns and ‘n’ signifies numerical columns.  In the above example, the dataTypeCode should be “innnni”.  The code tells vancova program that dataFile contains 6 columns and specifies the type of each column.

 

–c or –contrast contrastFile    

Specifies a text file containing the contrast being tested.  The numbers in contrastFile should correspond to the columns in the dataFile.  That is, one number for each column.  In the above example, the contrastFile may look like:

            0.0             0.0             1.0             –1.0             0.0             0.0

When a zero is specified in the contrast file, it means that the corresponding independent variable is controlled for or covaried out in the ANCOVA analyses.

 

Optional arguments:

 

–dataMask dataMaskCode

dataMaskCode is a string exclusively of characters ‘0’ or ‘1’.  The length of the string must equal the number of columns in the dataFile.  If a 0 is specified in a given position in dataMaskCode, the corresponding column is ignored and not used in the ANCOVA analyses.  The default value for this argument is “11..1”, that is, all codes are given the value of 1.  This means that by default, all columns will be used in the analyses.  For the above example, the dataMaskCode may be “101101”, which would indicate that the program should ignore columns 2 and 5 in the ANCOVA analyses.

 

–FWHM   FWHM

Specifies the full width at half maximum (in millimeters) of a Gaussian filter that is applied to all input images before analyses are performed.  Default FWHM=0.0.

 

–m or –mask   mask  

This option specifies an ANALYZE image of type ‘short’.  Voxels where the mask image value is zero are not used in the ANCOVA analyses. If no mask image is specified, all image voxels are analyzed.

 

–nozero

When this option is selected, voxels where the dependent variable takes on a value of zero are not analyzed, that is, the t-value is set to zero at those voxels.

 

–o or –output prefix   

This option can be used to specify a prefix for the output files.  The default value is prefix=V.

 

 

–v or –verbose          

Runs the program in verbose mode.

 

–x column  –y row  –z slice  

These options are intended for testing the software.  When specified, a table in created and written to the log file.  This table can be ported into other statistical programs (e.g., SPSS) in order to validate the results of the current program at the specified voxel.

 

Outputs:

The program outputs a log file (prefix.log), and two ANALYZE images: prefix_t.hdr(img), and prefix_df.hdr(img).  prefix_t is of type ‘float’ and stores the t-values computed from ANCOVA analyses and the specified contrast.  The prefix_df is of type ‘char’ and stores the voxelwise degrees of freedom of each ANCOVA analysis.  For the most part, prefix_df is a constant image since the degrees of freedom are not expected to change from one voxel to another.

 

Related programs:

threshold_tmap

 

Bugs:

            Please let me know if you find any!