NiftyReg_Loop_Bash

Running NiftyReg in for-loop in MacOS/Unix environment using BASH script.

Download NiftyReg

NiftyReg 1.3.9 (as of May 2016) https://sourceforge.net/projects/niftyreg/

Install NiftyReg

http://sourceforge.mirrorservice.org/n/ni/niftyreg/install_NiftyReg_macLinux.pdf

Overall, the instruction is fairly accurate. Except, the project will be installed into 'build' folder but not into 'install' folder as it mentioned by the instruction. Based on my experience, CMAKE 2.8.5 worked on both OSX9 (with Xcode6) and OSX11 (with Xcode7). (CMAKE 2.8.5 is GUI based. Xcode may or may not be required.) After 'generate', '$make' and '$make install' will be excuted in Terminal/Command Window. The newer version of CMAKE (such as 3.2) did not work.

I did not do the following part. But it is working.

            NIFTYREG_INSTALL=<path_to_your_niftyreg_install> (/Users/XYXY/niftyreg_install in the current example)
            export PATH=${PATH}:/${NIFTYREG_INSTALL}/bin
            export LD_LIBRARY_PATH=${LD_LIBRARY_PATH}:${NIFTYREG_INSTALL}/lib

CMAKE version 2.8 https://cmake.org/files/v2.8/

Another instruction

http://cmictig.cs.ucl.ac.uk/wiki/index.php/NiftyReg_install

My favorite work flow using MATLAB

            Image file -> MATLAB file -> Nifty file -> NiftyReg -> Nifty file -> MATLAB file

NIfTI_20140122 (as of May 2016): Nifty Image Tool on MATLAB

http://www.mathworks.com/matlabcentral/fileexchange/8797-tools-for-nifti-and-analyze-image

by Jimmy Shen

23 Oct 2005 (Updated 22 Jan 2014)

http://www.mathworks.com/matlabcentral/profile/authors/757722-jimmy-shen

MATLAB to Nii (2D)

%% Input %%

% IM: image data

% filename: file name

% Ref: reference/target image data

% Mask: Mask data on reference/target image

save_nii(IM, [filename '.nii']) % IM.nii

save_nii(Ref, 'Ref.nii') $Ref.nii

imagesc(Ref);

[Mask, x_ref_mask, y_ref_mask] = roipoly;

save_nii(Mask, 'Mask.nii') $Mask.nii
Nii to MATLAB (Stack of 2D images)

%% Input %%

% i1: image ID number

% x_ref, y_ref: points[x,y] on the reference image

% Deformed Coordinate System: IM_def.nii

% Warped Image: IM_warped.nii

%% Deformed Coordinate System %%

Def_nii = load_nii('IM_def.nii');

Y_def = Def_nii.img(:,:,1,1,1);

X_def = Def_nii.img(:,:,1,1,2);

%% Reference Coordinate System %%

[X_ref,Y_ref] = meshgrid([1:size(X_def,2)], [1:size(X_def,1)]);

%% Point by Point Transformation %%

x_def(:,i1) = interp2(X_ref,Y_ref,X_def,x_ref,y_ref);

y_def(:,i1) = interp2(X_ref,Y_ref,Y_def,x_ref,y_ref);

%% Warped Image %%

Warped_nii = load_nii('IM_warped.nii');

Warped_IM = Warped_nii.img;

Use NiftyReg_Loop.sh

There are XXX source/moving 2D images to be registered onto the reference/target 2D image around the specific area specified by a mask.

%% Modify the number of loops using text editor %%

e.g. 1 to 721

              for NUM in `seq 1 1 721`

%% Run NiftyReg_Loop.sh in Terminal/Command Window

              bash NiftyReg_Loop.sh

%% Input files (in the same folder) %%

Target Image: Ref0001.nii

Source Images: IM0001.nii to IM0XXX.nii

Mark: Mask0001.nii (0 and 1 on Ref0001.nii)

%% Output files %%

Affine Transformation: IM0XXX_aff.txt. Quick and dirty affine transformation (principal stretch, translation, shear, and rotation) with no deformable local control.

Warped Images: IM0XXX_warped.nii. Warped source images; Target Image ~ Warped Image (Hopefully)

Deformed Coordinate System: IM0XXX_def.nii

CPP file: IM0XXX_cpp.nii

About NiftyReg

Personally, my favorite deformable image registration software developed by a group of scientists at University College London, United Kingdom.

https://sourceforge.net/projects/niftyreg/

Reference

LICENSE

CONTACT

For any comment, please, feel free to contact Marc Modat ([email protected]). https://sourceforge.net/p/niftyreg/git/ci/master/tree/

  May 29th 2016
      Tatsuya J. Arai

araitatsuya / niftyreg_loop_bash Goto Github PK