README.md

Zerodose

A tool to assist in personalized abnormality investigation in combined FDG-PET/MRI imaging. Created by the department of Clinically Applied Artificial Intelligence at Copenhagen University Hospital

Installation

Note that a python3 installation is required for ZeroDose to work. You can install ZeroDose via pip from PyPI:

$ pip install zerodose

Usage

Note!

• All input images should be affinely registered to MNI 2009a Nonlinear Symmetric/Assymmetric space (1×1x1mm). Use zerodose pipeline if ZeroDose should do the registration.
• A brain mask is required to run ZeroDose - we can recommend HD-BET.

Run ZeroDose

Create an sbPET and abnormality map:

$ zerodose run -i mr.nii.gz -p pet.nii.gz -m brain_mask.nii.gz -os sb_pet.nii.gz -oa abn.nii.gz

Run pipeline

Identical to zerodose run but with registration (NiftyReg) to and from MNI space. (Registration may take several minutes depending on image dimensions)

$ zerodose pipeline -i mr.nii.gz -p pet.nii.gz -m brain_mask.nii.gz -os sb_pet.nii.gz -oa abn.nii.gz

Run individual steps

Synthesize raw baseline PET

$ zerodose syn -i mr.nii.gz -m brain_mask.nii.gz -o sb_pet_raw.nii.gz

Intensity normalize raw sbPET

$ zerodose norm -i mr.nii.gz -m brain_mask.nii.gz -o sb_pet_raw.nii.gz

Create abnormality map

$ zerodose abn -p pet.nii.gz -s sb_pet.nii.gz -m brain_mask.nii.gz -o abn.nii.gz

Please see the Command-line Reference for details.

Docker

• TODO

Hardware requirements

• TODO

Issues and contributing

Contributions are very welcome. If you encounter any problems, please file an issue along with a description.