Content-type: text/html Manpage of reorient

reorient

Section: User Commands (1)
Index Return to Main Contents

NAME

reorient - Reorients diffusion tensors for consistency with an image transformation.

SYNOPSIS

reorient -trans <transformation> [options]

DESCRIPTION

Reads diffusion tensor files (single or multiple) and an affine transformation, in the format output by FSL's flirt program, or a non-linear warp expressed as a displacement field. The displacement field can be either three separate analyze files for the x, y and z components or one nifti file in the format output by FSL's fnirt. The program outputs diffusion tensors reoriented for consistency with the transformation using either the preservation of principal directions (PPD) algorithm or finite strain (FS) reorientation; Alexander et al, IEEE Trans. Medical Imaging 20(11) 1131-1139 2001 describes both methods. By default the program reads and writes to and from the standard input and output streams. The usual camino command line options specify input and output types and redirections.

Note that reorient does not warp the image, but just rotates diffusion tensors and leaves them in the same position. To warp the image, use another program such as FSL's flirt program.

EXAMPLES

Reorient the diffusion tensors in DT image SubjectA_DT_S30.Bdouble using the PPD algorithm according to the transformation in Shear30.txt:

reorient -trans Shear30.txt < SubjectA_DT_S30.Bdouble > SubjectA_DT_S30_PPD.Bdouble

Now using FS instead:

reorient -trans Shear30.txt -reorientation fs < SubjectA_DT_S30.Bdouble > SubjectA_DT_S30_FS.Bdouble

We recommend using PPD always. FS is slightly faster, but less precise.

To perform reorientation for a non-linear transformation, the command format is similar:

reorient -trans Warp.nii -datadims 128 128 60 -voxeldims 1.7 1.7 2.3 < SubjectA_DT_Warp.Bdouble > SubjectA_DT_WarpPPD.Bdouble

We need to specify the image and voxel size on the command line for non-linear transformations. The above is tested only for nifti warp files output from FSL's fnirt using the --fout option. If the transformation is in three separate analyze images, the command is:

reorient -xtrans WarpX -ytrans WarpY -ztrans WarpZ -datadims 128 128 60 -voxeldims 1.7 1.7 2.3 < SubjectA_DT_Warp.Bdouble > SubjectA_DT_WarpPPD.Bdouble

Here is an example using flirt to warp images and reorient to do the tensor reorientation. Suppose we have DT images from two subjects, SubjectA and SubjectB, that we want to align. First create FA images and use flirt to compute the transformation between the images by registering FA maps:

# Make FA maps
fa < SubjectA_DT.Bdouble > SubjectA_FA.Bdouble
fa < SubjectB_DT.Bdouble > SubjectB_FA.Bdouble

# Convert to analyze format
cp SubjectA_FA.Bdouble SubjectA_FA.img
analyzeheader -voxeldims 1.7 1.7 2.5 -datadims 128 128 60 -datatype double > SubjectA_FA.hdr
cp SubjectB_FA.Bdouble SubjectB_FA.img
analyzeheader -voxeldims 1.7 1.7 2.5 -datadims 128 128 60 -datatype double > SubjectB_FA.hdr

# Run the registration
flirt -in /tmp/SubjectA_FA.img -ref SubjectB_FA.img -out SubjectA_FA_Warped.img -omat SASB_Affine.txt

Now use flirt to warp each component of the diffusion tensor image SubjectA_DT.Bdouble separately using the transformation SASB_Affine.txt.

for ((i=0; i<8; i=i+1)); do

    # Create volume of i-th component in analyze format
    shredder $((i*8)) 8 56 < SubjectA_DT.Bdouble > SubjectA_DT_Comp${i}.img;
    analyzeheader -voxeldims 1.7 1.7 2.5 -datadims 128 128 60 -datatype double > SubjectA_DT_Comp${i}.hdr

    # Transform.  Note that the first component contains exit codes, so
    # must be transformed using nearest neighbour interpolation. The others
    # are transformed using linear interpolation.
    if [ $i == 0 ]; then
        flirt -in SubjectA_DT_Comp${i}.img -applyxfm -init SASB_Affine.txt -out SubjectA_DT_Comp${i}_Warped.img -ref SubjectB_FA.img -interp nearestneighbour
    else
        flirt -in SubjectA_DT_Comp${i}.img -applyxfm -init /tmp/SASB_Affine.txt -out SubjectA_DT_Comp${i}_Warped.img -ref SubjectB_FA.img
    fi

done

# Recombine into a single DT file in camino format. On a linux machine,
# flirt outputs results with little-endian byte ordering, which we need
# to reverse.
cat SubjectA_DT_Comp?_Warped.img | shredder 0 -8 0 | scanner2voxel -voxels 983040 -components 8 -inputdatatype double -outputdatatype double > SubjectA_DT_WarpedNR.Bdouble

# Now reorient the tensors using PPD
reorient -trans /tmp/SASB_Affine.txt < SubjectA_DT_WarpedNR.Bdouble > SubjectA_DT_WarpedPPD.Bdouble

OPTIONS

-trans <Transformation File>

Specifies the file containing the transformation. It may be a text file containing an affine transformation. In that case, the format is that output by FSL's flirt program, which is a text file containing a 4x4 affine transformation matrix in the following format:

Lxx Lxy Lxz Tx
Lyx Lyy Lyz Ty
Lzx Lzy Lzz Tz
0 0 0 1

Or is may be a nifti file containing three components, x, y and z, of a displacement field. The format is that output by FSL's fnirt program using the --fout option.

-xtrans, -ytrans, -ztrans <Transformation File>

Specifies three components of a displacement field in separate analyze images. Equivalent to using -trans with a nifti file.

-reorientation <strategy>

Specifies the reorientation strategy. The program currently supports options fs or ppd; ppd is the default.

AUTHORS

Daniel Alexander <camino@cs.ucl.ac.uk>

BUGS

This document was created by man2html, using the manual pages.
Time: 02:07:11 GMT, December 04, 2017