-
Notifications
You must be signed in to change notification settings - Fork 3
Dataset
The dataset we use for this project is the "Bullitt ealthy MR Database" from kitware.com
This dataset contains 100 (T1, MRA) image pairs from healthy patients, acquired by a Siemens Allegra head-only 3T MR system. The voxel spacing for the MRA images is (0.5x0.5x0.8mm) with a volume size of (448x448x128voxel); The voxel spacing for the T1 images is 1mm isotropic with a volume size of (176x256x176voxel). The full dataset can be downloaded here. A subset of this dataset also includes intra-cranial vasculature (centerline + radius), extracted from the MRA images. These models are found in each patient's "Auxillary Data" folder, such as here.
You can have more information on the original page : https://public.kitware.com/Wiki/TubeTK/Data
You can downlowd the dataset on this page : https://data.kitware.com/#collection/591086ee8d777f16d01e0724/folder/58a372e38d777f0721a64dc6
Before beginning, you need to convert the image to NIFTI file format.
To convert the MRA from MHA to NIFTI format, you can use the script mha_to_nii.py located in the tools folder which is using SimpleITK.
python3 tools/mha_to_nii.py ../Datasets/Bullitt/ ../Datasets/Bullitt_NII/MRA/ -MRA.mha
To convert the ground truth from 'tre' to NIFTI format, you will need to use TubeTK. I did not do it myself so I can really help you with that.