Test data for Karawun

This repository contains a set of nifti files (in the Images folder) for conversion to Brainlab compatible DICOM images using Karawun (). Karawun users can use this dataset to satisfy themselves that nifti data is converted sensibly, independent of the data stride pattern in the input nifti file, or anisotopic slicing.

Image Data

The source data is the standard MNI atlas with left/right markings distributed with FSL.

Several different versions of this image have been created:

1. Reslicing to produce anisotropic voxels. For example thicker axial slices.

2. Reformatting data layout - i.e modifying the stride pattern

Masks or labelled versions of the MNI atlas have also been created that include the "L" and "R" markers only. Each letter is a separate connected component and will be displayed in a different colour in Brainlab.

These images were also resliced and reformatted.

Streamline data

A set of synthetic "tracts" have been created using mrtrix streamline files. The streamlines are best viewed in the axial direction and form a pair of letter ("L" and "R") and are aligned with the original markers.

Creating the data

See Images/README for commands used to create the images

Naming conventions

The images have name prefixes of either mni or LR, with mni indicating graylevel images and LR indicating mask images. Suffixes are of the form 2_2_4 indicating voxel dimensions and A through E indicating different stride patterns.

Details can be queried further using fslinfo or mrinfo. The images and tck files can also be viewed using mrview

Conversion

The test dataset can be converted to Brainlab compatible using the Karawun command:

mkdir brainlabdicom

importTractography -d ./Dicom/1.3.12.2.1107.5.2.43.167031.2019040213095021814319052.dcm --nifti Images/MNI152_T1_2mm_LR-masked.nii.gz mni_* --label-files Images/LR* --tract-files Images/letters.tck -o brainlabdicom

Verify

All data in this test set has been derived from a single image using steps that preserve geometry. i.e the location of any chosen landmark should be the same across all versions of the same image, and the position of the tracts and 3D objects should remain consistent across all combinations of image versions, irrespective of original stride pattern or slicing.