PETFit: A BIDS App for PET Kinetic Modelling

PETFit is a BIDS App for fitting kinetic models to PET time activity curve (TAC) data. It runs a configurable, step-by-step pipeline — each step accompanied by detailed HTML reports for quality control — and delegates the kinetic model fitting to kinfitr.

Full documentation: https://petfit.readthedocs.io

This README is only a brief overview. For the complete installation guide, tutorials, command reference, container/HPC usage, supported models, and troubleshooting, see the documentation site.

Note: PETFit is currently in active development, and there may be bugs. Please report them on the GitHub issues page — they are extremely valuable for making the pipeline robust for all datasets.

Overview

PETFit can be used in two ways:

• Interactive mode — graphical web apps to configure analyses step-by-step, run individual steps, and save configuration files for reproducible processing.
• Automatic mode — run the full pipeline non-interactively from a saved configuration file, locally or on a server/HPC.

It provides three apps for interactive usage:

• Region Definition — combines regional TACs from BIDS PET preprocessing derivatives.
• Modelling with Plasma Input — invasive models requiring a blood input function (1TCM, 2TCM, 2TCM_irr, Logan, MA1, Patlak).
• Modelling with Reference Tissue — non-invasive models using a reference region (SRTM, SRTM2, refLogan, MRTM1, MRTM2, refPatlak).

Installation

PETFit can be run as a container (Docker / Apptainer), or as an R package.

Docker:

docker pull mathesong/petfit:latest

The easiest way to drive the Docker image is the petfit-docker command-line wrapper. It turns a simple BIDS-App-style command into the matching docker run invocation, mapping your directories into the container for you:

pip install petfit-docker
petfit-docker --help   # see all options

R package:

# install.packages("remotes")
remotes::install_github("mathesong/petfit")

See the installation guide for full details, including Apptainer/HPC usage.

Quick start

A PETFit workflow has two stages: define regions once for the dataset, then run one or more modelling analyses. Launch each app interactively with the petfit-docker wrapper, then open http://localhost:3838 in your browser.

Region definition:

petfit-docker /path/to/bids /path/to/derivatives participant \
  --app regiondef \
  --cores 1

Modelling with plasma input:

petfit-docker /path/to/bids /path/to/derivatives participant \
  --app modelling_plasma \
  --blood-dir /path/to/blood \
  --cores 1

Modelling with reference tissue:

petfit-docker /path/to/bids /path/to/derivatives participant \
  --app modelling_ref \
  --cores 1

For the equivalent raw docker run commands (useful if you ever hit an issue with the wrapper), see the Docker usage guide.

For the full walkthrough — automatic processing, Apptainer/HPC, configuration, outputs, and troubleshooting — see the Quick start and Usage guides.

Citation

If you use PETFit in your research, please cite kinfitr for now:

An introduction to the package:

Matheson, G. J. (2019). Kinfitr: Reproducible PET Pharmacokinetic Modelling in R. bioRxiv: 755751. https://doi.org/10.1101/755751

A validation study compared against commercial software:

Tjerkaski, J., Cervenka, S., Farde, L., & Matheson, G. J. (2020). Kinfitr – an open source tool for reproducible PET modelling: Validation and evaluation of test-retest reliability. EJNMMI Res 10, 77 (2020). https://doi.org/10.1186/s13550-020-00664-8

Contributing

Contributions are welcome! Please report issues or submit pull requests on GitHub at https://github.com/mathesong/petfit.

License

This project is licensed under the MIT License — see the LICENSE file for details.