Arxiv • Usage • Installation • Cite

SynthBA (Synthetic Brain Age) is a deep-learning model able to predict the biological age of the brain (referred to as brain age) through brain MRIs of arbitrary contrast and resolution. It follows the idea of using domain-randomization from the seminal work of SynthSeg.

Running SynthBA requires docker (see the Docker installation page) and nothing else. Once docker is installed, you can run the latest version of SynthBA (see DockerHub) in one command:

./synthba --help

Place your MRIs (nifti format) in a folder (/path/to/inputs) and create a folder where to store the outputs (/path/to/outputs). Then run:

./synthba /path/to/inputs /path/to/outputs \
  -m <MODEL> -b <BATCHSIZE> -t <TEMPLATE>

Or directly using Docker:

docker run --rm \
    -v /path/to/inputs:/home/inputs \
    -v /path/to/outputs:/home/outputs \
    lemuelpuglisi/synthba:latest \
        -m <MODEL> -b <BATCHSIZE> -t <TEMPLATE>

Configure the batch size according to your available RAM. Templates are currently needed to align the scans to MNI space. We provide T1w and T2w templates, which should work for a wide range of MRI sequences. A list of <MODEL> and <TEMPLATE> options are provided below:

The output folder will contain both the predicted brain age for each input (predictions.csv) and the preprocessed scans (in preprocessed).

If you want to build the SynthBA's Docker image locally, run the following command at the root of the project:

docker build -t synthba .

Verify the installation by calling the --help option:

docker run --rm synthba --help

Cite the preprint:

@misc{puglisi2024synthba,
      title={SynthBA: Reliable Brain Age Estimation Across Multiple MRI Sequences and Resolutions}, 
      author={Lemuel Puglisi and Alessia Rondinella and Linda De Meo and Francesco Guarnera and Sebastiano Battiato and Daniele Ravì},
      year={2024},
      eprint={2406.00365},
      archivePrefix={arXiv},
      primaryClass={eess.IV}
}