PyPeT is an open-source Python-based tool for automated quantitative analysis of brain CT and MR perfusion imaging. It generates standard perfusion maps including cerebral blood flow (CBF), cerebral blood volume (CBV), mean transit time (MTT), time-to-peak (TTP), and time-to-maximum (Tmax) from raw 4D perfusion data.
Unlike many commercial tools, PyPeT is free, transparent, and designed with modularity and customizability in mind. It supports both CT Perfusion (CTP) and MR Perfusion (MRP) within a unified framework, making it broadly applicable in cerebrovascular research.
For a detailed description of the pipeline and the validation experiments, please read the attached technical report.
- Open and extensible: Python implementation with clear structure, strong modularity, and extensive documentation
- Low computational requirements: Runs in seconds on a standard laptop, no GPU needed
- Debug mode: Step-by-step visualization of preprocessing, arterial input function (AIF) selection, and deconvolution steps
- Validation: Outputs benchmarked against multiple FDA-approved software package, showing strong correspondence (SSIM ~0.75–0.85)
If you use this tool in your research, please cite
Borghouts M, Su R. PyPeT: A Python Perfusion Tool for Automated Quantitative Brain CT and MR Perfusion Analysis. arXiv
- Create and activate a venv
conda create --name pypet python==3.10
conda activate pypet
- Clone this repository
git clone https://github.com/Marijn311/CT-and-MR-Perfusion-Tool.git
- Install requirements
cd CT-and-MR-Perfusion-Tool
pip3 install -r requirements.txt
This tool expects input data to follow a specific directory structure. The expected structure is demonstrated in the demo_data folder.
Though with some tweaking of main.py you may use other dataset structures.
data/
├── sub-<subject_id>/
│ └── ses-<session_id>/
│ ├── sub-<subject_id>_ses-<session_id>_ctp.nii.gz # Raw CTP data (4D volume)
│ ├── brain_mask.nii.gz # (Optional) Brain mask
│ └── perfusion-maps/ # (Optional) Reference perfusion maps
│ ├── sub-<subject_id>_ses-<session_id>_cbf.nii.gz # Cerebral Blood Flow
│ ├── sub-<subject_id>_ses-<session_id>_cbv.nii.gz # Cerebral Blood Volume
│ ├── sub-<subject_id>_ses-<session_id>_mtt.nii.gz # Mean Transit Time
│ └── sub-<subject_id>_ses-<session_id>_tmax.nii.gz # Time to Maximum
You may provide your own brain mask, if no brain mask is provided, the tool will generate brain masks.
The PyPeT tool is configured through the config.py file, settings should be self-explanatory.
Additional parameters, such as the choice of deconvolution method or threshold values, can be adjusted in the arguments of the respective functions within the utils folder.
To process a dataset, run the main.py script. This will automatically generate perfusion maps for all inputted perfusion scans, and, if reference maps are available, compare the outputs against them. For single-image processing, you can instead call the core() function directly to generate perfusion maps from one input file.