Exotic DICOMs (Work in Progress!)

Note that this repository is a work in progress. We are actively looking for interesting DICOMs to add to this repository. If you have any interesting DICOMs that you would like to share, please open a pull request with the DICOMs and a README.md file that describes the DICOMs.

Exotic DICOMs provides a simple way to test DICOM conversion functions in python, and beyond. It contains a wide range of DICOMs to check if your DICOM conversion functions can handle a wide-range of edge-cases

Quickstart

pip install git+https://github.com/radoss-org/exotic-dicoms.git

This is a quickstart on testing dicoms with RadStract (https://github.com/radoss-org/radstract)

import pydicom
from radstract.data.dicom import convert_dicom_to_images

from exotic_dicoms.test import test_dicom_image_conversion


def pre_step(dcm):
    return pydicom.dcmread(dcm)

# This will run the convert_dicom_to_images function for each DICOM in the test data.
# and check if the output is correct.
results = test_dicom_image_conversion(
    func=convert_dicom_to_images,
    dicom_kwarg_name="old_dicom",
    prestep=pre_step,
)

Since radstract only accepts pydicom objects as input, we use a prestep function to convert the path that exotic_dicoms provides to a pydicom object.

Viewing bad DICOMs

If you get bad results, here is how you can inspect the bad DICOMs:

import pydicom
from radstract.data.dicom import convert_dicom_to_images

from exotic_dicoms.test import test_dicom_image_conversion


def pre_step(dcm):
    return pydicom.dcmread(dcm)


results = test_dicom_image_conversion(
    func=convert_dicom_to_images,
    dicom_kwarg_name="old_dicom",
    prestep=pre_step,
)

for filename, bad_results in results:
    for i, bad_result in enumerate(bad_results):
        img = bad_result.visualize()
        img.save(f"bad_{filename}_{i}.png")

Manual Checking

You can download the test cases directly from the following link:

https://github.com/radoss-org/exotic-dicoms/tree/main/exotic_dicoms/test-data

You should see this exact data from each DICOM: https://files.mcaq.me/8086j.mp4

License

This project is licensed under Apache 2.0. See the LICENSE file for more details.

Call for Contributions

We are actively looking for interesting DICOMs to add to this repository. If you have any interesting DICOMs that you would like to share, please open a pull request with the DICOMs and a README.md file that describes the DICOMs.