Header

15. Introduction to DRTAMAS

DRTAMAS is a a tensor based registration software, released under the TORTOISE processing package.This module of processing is mainly used to perform tensor based registration or template creation, using diffusion tensors (DTs) created by the TORTOISE processing pipeline. While running any of these commands from the bin folder, shown below, it is advised that you copy all the DTs that will be used in the registration, the text file to be used, and any reference template, into a separate folder. This keeps the template creation or any DRTAMAS generated results separate from rest of the data processing.

This page will focus on the main commands that are used in the template creation and registration of tensors. Other commands, than the ones mentioned here, are used internally as part of these main commands.

15.1. DTIREG

DTIREG command should be used if you want to register one diffusion tensor to another.

15.2. DTI CREATE TEMPLATE

DTICreateTemplate command performs the template creation, from the list of (DTs) provided in the text file. The syntax to run the command is as follows:

If you would like to create a group template from the all the diffusion tensors in your study or specific group, this command will generate the average tensor. The final average tensor that comes out of this process is named as average_template_diffeo_6.nii In the process, the tensors that were used in the creation of this template, will also be warped to the final template and will be in the same folder with the subject name.diffeo.nii

It is important therefore to make sure the individual images have unique names, before you begin processing.

Note: The quality of the template is as good as the input images provided. So, please make sure the prior steps listed in this documentation have been performed.

15.3. DTI batch register to target

When using this command, please copy the text file to be used and the target DT into a separate folder.

 

 

EXAMPLES:

1) Creation of a population template 

Requirements: tortoise generated DT files, a text file containing the full path of all DT files that will go in the creation of the template i.e DT.txt

We will create a specific folder, called population template. This folder mainly contains the DT.txt file and will contains all the resulting outputs from template creation. If, for example, you want to create a template out of three subjects, the DT.txt file will have contents as follows: 

/folder1/subj1_dt.txt

/folder2/subj2_dt.txt

/folder3/subj3_dt.txt

 dtireg_create_template command will be run as follows:

dtireg_create_template   -s= ~/population_template/DT.txt    -n=1  -c=1  -r=1  

In this example command, we are using one CPU with n=1, constraining the deformation c=1, and the resulting output resolution of the final template will be 1mm, r=1. You may also provide an initial rigid template and/or the final output folder. 

15.4. outputs of create template or DTIREGcreate_template_output

 

Here is an example folder of outputs generated for dti_create template. This folder consists of the .txt file with the original DTs “control_list.txt”.

The command runs by first picking the rigid template “rigid_template_0.nii”, representative of the group. Then the command runs to iteratively output affine files and the diffeo files. The final template DT is the “average_template_diffeo_6.nii”. As you can also see, there are additional files output for the corresponding subject files.

Aff.txt= is the text file comprising of the affine transformation. aff.nii = the native files warped to the template space, after the application of the aff.txt file. diffeo.nii = the native DT transformed to the final template space.

diffeo_proc = is not computed automatically but can be done by either running the commands such as compute FA or compute TR listed under the DRTAMAS folder. Alternatively, to derive more scalar variables from the warped DT, you may import the diffeo.nii file into TORTOISE GUI and derive variables. If you are using the TORTOISE 2.5.2 GUI to derive variables, use the following steps: Import diffeo.nii into TORTOISE using the import Tensors button in the GUI (TORTOISE 2.5.2 DIFFCALC). Click on compute eigen values, then click on the derive variables, when the button gets activated. This should take a few minutes depending on the resolution of the image. The ROI utilities will then get active and click on this button to reveal all the scalar maps derived from your warped tensors. Please close the gui after viewing your images and then make sure to click on the save session button. This will save the session with an extension _L0_SAVE under _proc folder.

 

2) Registering subjects in a population to a pre-existing template or template created from dti_create_template

If you have a template (ex: control or population) TORTOISE average tensor, you can batch register your individual subject tensors to the average using the command:

dti_reg_batch_register_to_target

This command takes a list of individual DT's and registers them to the target DT provided. The requirements for this task are:

a) a text file with list of full path to subject DT's. The organization of the text file you provide to create the template. 

b) a template DT file to which you want all the subjects registered

*pro tip: highly recommend creating a separate folder and copying the template DT file and the saving the text file with the list of DT's, that you want to register to the template. In this case the new folder that we will create will be: individual_registration_to_template

With (a) and (b) you  can run the command as follows:

dti_reg_batch_register_to_target     -t=/individual_registration_to_template/average_template_DT.nii     -s=/individual_registration_to_template/DT.txt     -n=1    -z=1   

 

15.5. Combine Transformations

* The steps noted under this sections mainly applies to generating combined deformation files from the outputs generated for dti_create_template.

** Please note: The dti_reg_batch_register_to_target will output combined transformations (a.k.a _aff_diffeo_MINV.nii) files and you will not need to do the steps explained below. 

Here you will combine the files ending in “aff.txt” (scalar transformation) and “deffield_MINV.nii” (higher order transformation) to create the combined transformation. The transformation file gives information of deformation field generated, from warping the individual DT to the final template space. Please remember that diffeo.nii file output is the warped native DT into the final average template space.

You have three types of combine transformations command, each with same type of output but provisions for customizing the naming convention and image parameters.

If you use combineTransformations: The output file will be a generic name, with combineddeformation.nii as the output. If you have multiple subjects in a directly, these will be overwritten. This command can only be used if you have one set of deformation fields you want to combine in a folder.

If you have multiple subjects and correspoding transformations to be combined, please use CombineTransformationsWithOputputName command. This will give you flexibility to name the output file with specific subject names.

If you would like to output the transformations files with the imaging parameters similar to a reference DT, please use the command: CombineTransformationsWithReferenceImage. This will ensure the origin information from the reference image is translated to the transformation file being created.

 

15.6. Apply transformations to scalar

The transformation file created in the combined transformation for each subject, can be applied to the native, non warped scalar map from the corresponding subject to bring it into the final template space. This is especially useful if you like to warp each subject’s anatomical map into the final template space.

15.7. Apply transformations to tensor

The transformation file created in the combined transformation for each subject, can be applied to the native, non warped tensor map from the corresponding subject to bring it into the final template space. This maybe advantageuos in scenarios where you have created an alternate tensor map such as with dual compartment fitting in TORTOISE and would like to warp those tensors into final template space.

15.8. Apply transformations to 4d image

You may apply the transformations to any 4d image file to bring it into final template space.

Note: Please cite the following paper if you use this tool: DR-TAMAS: Diffeomorphic Registration for Tensor Accurate Alignment of Anatomical Structures, M. Okan Irfanoglu, Amritha Nayak, Jeffrey Jenkins,Elizabeth B.Hutchinson,Neda Sadeghi,Cibu P.Thomas,Carlo Pierpaoli,NeuroImage,Volume 132, 15 May 2016, Pages 439-454

Previous PostNext Post