This page intends to provide an overview of all tools that are included with the current MITK Diffusion Installer and refer to the respective Plugin in the MITK Diffusion application (if one exists). For a detailed list of parameters call the according tool without any arguments (see MITK MiniApps for details on this) or refer its Plugin equivalent.

Preprocessing Tools

Diffusion Dicom Loader

Loads Diffusion DICOM Files and creates NRRD or Nifty file.

$./MitkDiffusionDICOMLoader -i /home/dcmFolder/ -o /home/outputfile.nrrd

$./MitkDiffusionDICOMLoader -i /home/dcmFolder/ -o /home/outputfile.nii

Regular Dicom Loader

Loads DICOM Files and creates NRRD or Nifty file.

$./MitkDicom2Nrrd -i /home/dcmFolder/ -o /home/outputfile.nrrd

$./MitkDicom2Nrrd -i /home/dcmFolder/ -o /home/outputfile.nii

Folder Registration

Allows to register a series of images (of different modalities, including diffusion weighted) to one reference image. It allows to register derived resources (e.g. a segmentation on a T2 image) using the transformation of the original (T2) image.

For the following examples assume a folder containing a longitudinal study with T1,T2, DWI images and segmentations (ROI) :

Patien01_2010-1.dwi
Patien01_2010-1_T1.nrrd
Patien01_2010-1_T2.nrrd
Patien01_2010-1_ROI.nrrd
Patien01_2010-2.dwi
Patien01_2010-2_T1.nrrd
Patien01_2010-2_T2.nrrd
Patien01_2010-2_ROI.nrrd
Patien01_2010-3.dwi
Patien01_2010-3_T1.nrrd
Patien01_2010-3_T2.nrrd
Patien01_2010-3_ROI.nrrd
Patien01_2010-4.dwi
Patien01_2010-4_T1.nrrd
Patien01_2010-4_T2.nrrd
Patien01_2010-4_ROI.nrrd

All T2 and DWI images are to be co-registered to the first T2 image, this can be achieved by the following two calls:

$./MitkRegistration -i /home/inputFolder/ -o /home/outputFolder/ -f Patien01_2010-1_T2.nrrd -m T2.nrrd

$./MitkRegistration -i /home/inputFolder/ -o /home/outputFolder/ -f Patien01_2010-1_T2.nrrd -m .dwi

The segmentations where performed on the T1 image and are therefore related to the image space of the respective T1 image, so they can be bound to these images by marking them as derived resources. To register them both you would call

$./MitkFolderRegistration -i /home/inputFolder/ -o /home/outputFolder/ -f Patien01_2010-1_T2.nrrd -m _T1.nrrd -d _ROI.nrrd -b

Note: the suffixes of '_T1.nrrd' and '_ROI.nrrd' must have the same length!

The parameter -b designates the derived resource as binary such that a nearest neighbor interpolation is used.

All images (execpt for DWI files) are resample to the reference image, to resample to a specific spacing append the desired spacing like this (e.g. 1 x 1 x 2 mm)

$./MitkFolderRegistration -i /home/inputFolder/ -o /home/outputFolder/ -f Patien01_2010-1_T2.nrrd -m .dwi -r 1,1,2

Note: Registration methods assume that both images occupy roughly the same space. It may happend that this is not the case, and therefore registration fails. In this case you can try the -c option which uses the same origin for both images.

Copy Geometry

Copies the geometry (origin) of the source image to the target image. This is useful in cases where rounding errors in geometries cause the following itk error: itk::ERROR: ImageToImageFilter(): Inputs do not occupy the same physical space!

For images with differing extent the –alignCentroid option will adapt the origin of the input image such that the centroids of both images align, this can be as a preprocessing steo for registratering images that do not overlap at all (e.g. error : Joint PDF summed to zero )

Resampling of Images

This tool can be used to resample images in different ways. One way is by specifying the spacing. Transforming an image test.nrrd to have a spacing of 1x1x3 mm can be done using this command:

$./MitkImageResampler -i test.nrrd -s 1,1,3 -o resampledImage.nrrd

Alternatively this program can be used to resample by reference. This will resample the input image to the grid provided by the reference image and also ensure they occupy the same physical space (that is the input image will have the same origin and voxel dimensions)

$./MitkImageResampler -i test.nrrd -r referenceImage.nrrd -o resampledImage.nrrd

Note: Resampling by reference image can also be applied to MR Diffusion data (.dwi files).

Tensor Reconstruction

See Tensor Reconstruction for the GUI equivalent of this tool.

Takes a .dwi, .fsl/.fslgz file as input and saves the computed reconstructed tensor to the specified file. It also allows for a threshold to be set, to exclude low b values from the reconstruction process.

$./MitkTensorReconstruction -i /home/user/sample.dwi -o /home/user/tensors.dti -t 50

Qball Reconstruction

See Q-Ball Reconstruction for the GUI equivalent of this tool.

./MitkQballReconstruction -i /home/user/sample.dwi -o /home/user/tensors.qbi -t 50 -r .006 -shc /home/user/coeffs.csv

Peak Extraction

Extracts ODF peaks from the given spherical harmonics coefficient image. Input image type is an image that contains a vector with the spherical harmonics coefficients as pixel type: Image< Vector< float, (ShOrder*ShOrder + ShOrder + 2)/2 + ShOrder >, 3 >

Peak Angular Error

Calculates the angular error between two sets of input directions. The directions are stored as images. Each image voxel contains one direction vector. Such images are for example the output of the fiber direction extraction miniapp.

Diffusion Related Measures

Diffusion Indices

See QmitkDiffusionImagingUserManualQuantification for the GUI equivalent of this tool.

Computes a selected tensor derived indices (fa, gfa, ra, ad, rd, ca, l2, l3, md) given a Tensor, Q-ball or FSL/MRTrix SH-coefficient image. E.g. to compute the fraction anisotropy call

./MitkDiffusionIndices -i /home/user/input.dti -idx fa -o /home/user/fa_image.nrrd

Tensor Derived Maps Extraction

Similar to Diffusion Indices . But computes all of the following indices FA, RA, MD, CA, RD, AD at once. Also the input is a regular .dwi file, the tensor reconstruction is done implicitly (using a b0 threshold of 50).

Fiber Tracking and Processing Methods

Fiber Direction Extraction

Extracts the voxel-wise main fiber directions from a tractogram.

Streamline Tracking

See Deterministic Streamline Tractography for the GUI equivalent of this tool.

Performs streamline tractography on a tensor image.

Gibbs Fiber Tracking

See Global Gibbs Tractography for the GUI equivalent of this tool.

Performs global Gibbs tractography on a tensor/Q-ball/SH-coefficient image.

Fiber Processing

Post-process a fiber bundle. Provides the possibility to

remove short/long fiber tracks
apply curvature threshold
resample a fiber bundle (linear and spline based)
compress a fiber bundle (lossy)
transform fiber bundle (scale, translate, rotate)
mirror fiber bundle

Fiberfox

See Signal Generation for the GUI equivalent of this tool.

Generates a signal from a fiber bundle provided a reference DWI and a parameter file. The parameter file can be generated using the Fiberfox plugin (sub-tab) Signal Generation.

File Format Converter

Determines the data type and converts the input file (if possible) to .NRRD (regular image), .DWI (diffusion image) or .FIB (fiber bundle).

Multishell Methods

Computes several fits on an images (Kurtosis,Bi-Exponential, ADC).

These fits are part of the Preprocessing Plugin QmitkDiffusionImagingUserManualPreprocessing .

Connectomics

Network Creation

See The Connectomics Network Data View for the GUI equivalent of this tool.

Creates a network based on a brain parcellation and a fiber image.

Network Statistics

See The Connectomics Statistics View for the GUI equivalent of this tool.

Calculates several network statistics for a given connectome.

Table of Contents