Installation

python -m pip install phantominator

The goal is to have easy installation and usage for everyone. If something doesn't work, please open an issue and/or submit a pull request. We'll get it figured out.

pygrappa is an optional dependency required to run the phantominator.examples.radial_coil_sens example.

About

Python package for easy generation of numerical phantoms. I often need a simple image to try something out on. In MATLAB, I would use the phantom command to quickly get something to work with. In Python, it's not always quite so easy, so I made this package that's quick to install and use so there's as little friction as possible. There are other implementations of Shepp-Logan available from other projects, but they are usually not as easy to install or include other things that I don't want for this project.

This package offers both CT and MR versions.

Going forward, this module will support Python >= 3.8.

Usage

Also see the examples module and docstrings. The interface for CT phantom generation is similar to MATLAB's phantom function.

Examples can be run as:

# python -m phantominator.examples.[example-name], e.g.:
python -m phantominator.examples.shepp_logan

Basic usage:

# CT phantom
from phantominator import shepp_logan
ph = shepp_logan(128)

# MR phantom (returns proton density, T1, and T2 maps)
M0, T1, T2 = shepp_logan((128, 128, 20), MR=True)

The Shepp-Logan 3D phantom has ellipsoids in [-1, 1] along the z-axis. The 2D Shepp-Logan exists at z=-0.25, so if we want just a subset along the z-axis with the first slice being the traditional 2D phantom, we can use the zlims option:

from phantominator import shepp_logan
M0, T1, T2 = shepp_logan((64, 64, 5), MR=True, zlims=(-.25, .25))

We can also generate simple oscillating concentric circles:

# Dynamic (concentric circles), 20 time frames
from phantominator import dynamic
ph = dynamic(128, 20)

If we want to modify ellipse/ellipsoid parameters or we just want to see what they are. For example, we can get the MR ellipsoid parameters like this:

from phantominator import mr_ellipsoid_parameters
E = mr_ellipsoid_parameters()

See docstrings for ct_shepp_logan, and mr_shepp_logan for how the array E is structured. It follows conventions from MATLAB's phantom function.

Arbitrary k-space sampling is supported for the single coil 2D Shepp-Logan phantom:

# Given k-space coordinates (kx, ky), where kx and ky are 1D
# arrays using the same unit conventions as BART's traj command,
# we can find the corresponding k-space measurements:
from phantominator import kspace_shepp_logan
k = kspace_shepp_logan(kx, ky)