Skip to main content

Python package for DICOM-SEG medical segmentation file reading and writing

Project description

pydicom-seg

License: MIT Python versions PyPI version DOI

Reading and writing of DICOM-SEG medical image segmentation storage files using pydicom as DICOM serialization/deserialization library. For detailed explanations about supported functionality and usage please have a look at the documentation.

Motivation

Converting DICOM-SEG files into ITK compatible data formats, commonly used for research, is made possible by the dcmqi project for some time. However, the project is written in C++ and offers only access to the conversion via the binaries itkimage2segimage and segimage2itkimage. After a conversion of a DICOM-SEG file to ITK NRRD file format, the user has to scan the output directory for generated files, load them individually and potentially combine multiple files to the desired format.

This library aims to make this process much easier, by providing a Python native implementation of reading and writing functionality with support for numpy and SimpleITK. Additionally, common use cases like loading multi-class segmentations are supported out-of-the-box.

Installation

Install from PyPI

pip install pydicom-seg

Install from source

This package uses Poetry (version >= 1.0.5) as build system.

git clone \
    --recurse-submodules \
    https://github.com/razorx89/pydicom-seg.git
cd pydicom-seg
poetry build
pip install dist/pydicom_seg-<version>-py3-none-any.whl

Development

After cloning the repository, please install the git pre-commit hook to enforce code style and run static code analysis on every git commit.

git clone \
    --recurse-submodules \
    https://github.com/razorx89/pydicom-seg.git
cd pydicom-seg
poetry install
poetry run pre-commit install

Getting Started

Loading binary segments

import pydicom
import pydicom_seg
import SimpleITK as sitk

dcm = pydicom.dcmread('segmentation.dcm')

reader = pydicom_seg.SegmentReader()
result = reader.read(dcm)

for segment_number in result.available_segments:
    image_data = result.segment_data(segment_number)  # directly available
    image = result.segment_image(segment_number)  # lazy construction
    sitk.WriteImage(image, f'/tmp/segmentation-{segment_number}.nrrd', True)

Loading a multi-class segmentation

dcm = pydicom.dcmread('segmentation.dcm')

reader = pydicom_seg.MultiClassReader()
result = reader.read(dcm)

image_data = result.data  # directly available
image = result.image  # lazy construction
sitk.WriteImage(image, '/tmp/segmentation.nrrd', True)

Saving a multi-class segmentation

Please generate a metainfo.json for the segments you want to serialize using the web-based editor from dcmqi.

segmentation: SimpleITK.Image = ...  # A segmentation image with integer data type
                                     # and a single component per voxel
dicom_series_paths = [...]  # Paths to an imaging series related to the segmentation
source_images = [
    pydicom.dcmread(x, stop_before_pixels=True)
    for x in dicom_series_paths
]
template = pydicom_seg.template.from_dcmqi_metainfo('metainfo.json')
writer = pydicom_seg.MultiClassWriter(
    template=template,
    inplane_cropping=False,  # Crop image slices to the minimum bounding box on
                             # x and y axes. Maybe not supported by other frameworks.
    skip_empty_slices=True,  # Don't encode slices with only zeros
    skip_missing_segment=False,  # If a segment definition is missing in the
                                 # template, then raise an error instead of
                                 # skipping it.
)
dcm = writer.write(segmentation, source_images)
dcm.save_as('segmentation.dcm')

License

pydicom-seg is distributed under the MIT license.

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

pydicom-seg-0.4.1.tar.gz (24.3 kB view hashes)

Uploaded Source

Built Distribution

pydicom_seg-0.4.1-py3-none-any.whl (27.7 kB view hashes)

Uploaded 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