Skip to main content

A torchio extension for shape processing

Project description

Medical Shape

UnitTest Docker Docker Stable Build Package PyPI pre-commit.ci status

A torchio extension for shapes and their processing.

Usage

medical_shape provides 3 major classes for usage:

Shape

Shape is a subclass from torchio.data.Image. It stores arbitrary pointclouds together with their descriptions and an associated affine matrix. The pointclouds are stored as 2D torch.Tensor in the form NxD where N is the number of points and D is the dimensionality of points (usually D=3).

ShapeSupportSubject

The ShapeSupportSubject is an extension of torchio.data.Subject to allow the inclusion of Shape-type objects into the subject. It should be used instead of torchio.data.Subject, whenever a shape is included and is also safe to use without a shape (will behave exactly like torchio.data.Subject in that case) as shapes often require special handling.

TransformShapeValidationMixin

This class is a transformation mixin to allow checks whether shape-support is required and to raise warnings if the incorrect Subject-type was used. All transformations supporting shapes should inherit from it (and as it is derived from torchio.transforms.Transform, it is also safe to use this class as a standalone baseclass).

Installation

This project can be installed either from PyPI or by cloning the repository from GitHub.

For an install of published packages, use the command

    pip install medical-shape

To install from the (cloned) repository, use the command

    pip install PATH/TO/medical-shape

You can also add -e to the command to make an editable install in case you want to modify the code.

You can also install the package directly from GitHub by running

    pip install git+https://github.com/justusschock/medical-shape.git

Docker Images

We provide a docker image for easy usage of the package and as a base image for other projects.

The file for this image can be found at dockers/Dockerfile. We provide both, a CPU-only and a CUDA-enabled image based on the NVIDIA NGC PyTorch image. These images can be found on DockerHub.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

medical_shape-0.1.0.tar.gz (15.1 kB view hashes)

Uploaded Source

Built Distribution

medical_shape-0.1.0-py2.py3-none-any.whl (22.4 kB view hashes)

Uploaded Python 2 Python 3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page