This repository contains the code for Brain image segmentation of White Matter (WM), Grey matter (GM) and Cerebrospinal fluid (CSF) in the IBSR_18 dataset.
- Create a conda environment
conda create -n ctreg python==3.9.13 anaconda -y && conda activate ctreg
- Install the requirements
pip install -r requirements.txt
First we create the metadata of the IBSR_18 dataset necessary to run the algorithm.Run the following line to perform this step.
python main.py
Preprocessing of the dataset that consists in Bias Field correction and Normalization between 0 and 255. Run the following line to perform this step.
python -m preprocessing.preprocessing
Registration of the training dataset into Validation or Test to build the multi-atlases.
Run the batchfilecreator.py to create the elastix batch files to register all the brains into the Validation or Test
sets. For example, run the following line of code for Validation.
python -m database.batchfilecreator --batch_type elastix -set_option Validation --parameter Par0009
That outputs 5 batch files that you should run before going to the next section in a folder that is created in
cwd()\elastix\NEW_FOLDER\elastix
Run the batchfilecreator.py to create the elastix batch files to run transformix and register all the labels into
the Validation or Test sets. For example, run the following line of code for Validation.
python -m database.batchfilecreator --batch_type transformix -set_option Validation --parameter Par0009
That outputs 5 batch files that you should run before going to the next section and perform the final segmentation, the
batch files are located in a folder that is created in cwd()\elastix\NEW_FOLDER\transformix
All the results coming from elastix are saved in a folder located in cwd()\data\BFC_registration\Par0009
To perform the final segmentation from the multi-atlas two strategies are proposed.
Run the following line of code to perform the segmentation by using hard majority voting on the Validation.
python -m segmentation.multiatlas_segmentation --parameter_folder Par0009 --registration_folder BFC_registration
That will output a *.csv file located in cwd()\metrics with the 3 different evaluation metrics per tissue DSC,
HD and RAVD.
Run the following line of code to perform the segmentation by using hard majority voting on the Test if all the previous are run to build the multi-atlas for the Test set from the Training set.
python -m segmentation.multiatlas_segmentation --parameter_folder Par0009 --test_boolean True
--registration_folder BFC_registration
That will output *.nii.gz files with the final segmentations per patient in a folder call cwd()\Final_segmentations
Run the following line of code to perform the segmentation by using weighted majority voting on the Validation.
python -m segmentation.multiatlas_segmentation_weightedvoting
That will output a *.csv file located in cwd()\metrics with the 3 different evaluation metrics per tissue DSC,
HD and RAVD.