dicomtrolley 2.0.1

Retrieve medical images via WADO, MINT, RAD69 and DICOM-QR

dicomtrolley

Retrieve medical images via WADO, MINT, RAD69 and DICOM-QR

• Requires python 3.7, 3.8 or 3.9
• Uses pydicom and pynetdicom. Images and query results are pydicom.Dataset instances
• Multi-threaded downloading using requests-futures

Installation

pip install dicomtrolley

Usage

Basic example

# Create a logged-in http session
session = VitreaConnection(
    "https://server/login").log_in(user,password,realm)
                           
# Use this session to create a trolley using MINT and WADO
trolley = Trolley(searcher=Mint(session, "https://server/mint"),
                  wado=Wado(session, "https://server/wado"]))

# find some studies (using MINT)
studies = trolley.find_studies(Query(PatientName='B*'))  

# download the fist one (using WADO)
trolley.download(studies[0], output_dir='/tmp/trolley')

Finding studies

studies = trolley.find_studies(Query(PatientName='B*'))

Basic query parameters can be found in core.Query. Valid include fields (which information gets sent back) can be found in fields.py:

studies = trolley.find_studies(
    Query(modalitiesInStudy='CT*', 
               patientSex="F", 
               min_study_date=datetime(year=2015, month=3, day=1),
               max_study_date=datetime(year=2020, month=3, day=1),
               include_fields=['PatientBirthDate', 'SOPClassesInStudy']))

Finding series and instance details

To include series and instance level information as well, use the queryLevel parameter

studies = trolley.find_studies(  # find studies series and instances
    Query(studyInstanceID='B*', 
          query_level=QueryLevels.INSTANCE)

a_series = studies.series[0]  # studies now contain series    
an_instance = a_series.instances[0]  # and series contain instances

Downloading data

Any study, series or instance can be downloaded

studies = trolley.find_studies(Query(PatientName='B*',
                                     query_level=QueryLevels.INSTANCE))

path = '/tmp/trolley'
trolley.download(studies, path)                             # all studies
trolley.download(studies[0]), path                          # a single study
trolley.download(studies[0].series[0], path)                # a single series
trolley.download(studies[0].series[0].instances[:3], path)  # first 3 instances

More control over download: obtain pydicom.Dataset instances directly

studies = trolley.find_studies(              # find study including instances
    Query(PatientID='1234', 
          query_level=QueryLevels.INSTANCE)

for ds in trolley.get_dataset(studies):      # obtain Dataset for each instance
    ds.save_as(f'/tmp/{ds.SOPInstanceUID}.dcm')

Multi-threaded downloading

trolley.download(studies, path, 
                 use_async=True,  # enable multi-threaded downloading 
                 max_workers=4)   # optionally set number of concurrent workers
                                  # defaults to None which lets python decide

Using WADO only, without search

from dicomtrolley.wado import Wado
from dicomtrolley.core import InstanceReference

instance = InstanceReference(series_instance_uid='1.2.1', study_instance_uid='1.2.2', sop_instance_uid='1.2.3')

wado = Wado(session, wado_url)
for ds in wado.datasets([instance]):
  ds.save_as(f'/tmp/{ds.SOPInstanceUID}.dcm')

DICOM-QR

Trolley can use DICOM-QR instead of MINT as a search method. See dicom_qr.DICOMQuery for query details.

dicom_qr = DICOMQR(host,port,aet,aec)
trolley = Trolley(searcher=dicom_qr, downloader=wado)

# Finding is similar to MINT, but a DICOMQuery is used instead
trolley.find_studies(  
    query=DICOMQuery(PatientName="BAL*",   
                     min_study_date=datetime(year=2015, month=3, day=1),
                     max_study_date=datetime(year=2015, month=4, day=1),
                     include_fields=["PatientBirthDate", "SOPClassesInStudy"],
                     query_level=QueryRetrieveLevels.STUDY)) 

RAD69

The RAD69 protocol is an alternative to wado for downloading DICOM images.

dicom_qr = DICOMQR(host,port,aet,aec)
trolley = Trolley(searcher=dicom_qr, 
                  downloader=Rad69(session=session,
                                   url="https://server/rad69"))

studies = trolley.find_studies(Query(PatientName="AB*"))
trolley.download(studies[0], path)  # rad69 download works exactly like wado 
trolley.download(studies[1], path,  
                 use_async=True)    # multi-threaded download is supported

Download format

By default, trolley writes downloads to disk as StudyID/SeriesID/InstanceID, sorting files into separate study and series folders. You can change this by passing a DICOMDiskStorage instance to trolley:

from dicomtrolley.storage import FlatStorageDir

#  Creates no sub-folders, just write to single flat file
storage = FlatStorageDir(path=tmpdir)
trolley = Trolley(searcher=mint, downloader=wado,
                  storage=storage)

You can create your own custom storage method by subclassing storage.DICOMDiskStorage:

from dicomtrolley.storage import DICOMDiskStorage

class MyStorage(DICOMDiskStorage):
  """Saves to unique uid filename"""
  def save(self, dataset, path):    
    dataset.save_as(Path(path) / uuid.uuid4())

trolley = Trolley(searcher=mint, downloader=wado,
                  storage=MyStorage())

DICOM Query types

For most DICOM queries you can use a Query instance:

from dicomtrolley.core import QueryLevels 
trolley.find_studies(Query(PatientID='1234', 
                           query_level=QueryLevels.INSTANCE)

If you want to have more control over backend-specific options you can use a backend-specific query like MintQuery or DICOMQuery. The query then needs to match the backend:

trolley = Trolley(searcher=Mint(session, "https://server/mint"),
                  wado=Wado(session, "https://server/wado"]))
trolley.find_studies(MintQuery(PatientID='1234', limit=5))

Examples

• Search for studies in MINT
• Search for studies in DICOM-QR
• Find and download studies
• Using WADO only
• Download studies with rad69

Alternatives

• dicomweb-client - Active library supporting QIDO-RS, WADO-RS and STOW-RS.
• pynetdicom - dicomtrolley's DICOM-QR support is based on pynetdicom. Pynetdicom supports a broad range of DICOM networking interactions and can be used as a stand alone application.

Caveats

Dicomtrolley has been developed for and tested on a Vitrea Connection 8.2.0.1 system. This claims to be consistent with WADO and MINT 1.2 interfaces, but does not implement all parts of these standards.

Certain query parameter values and restraints might be specific to Vitrea Connection 8.2.0.1. For example, the exact list of DICOM elements that can be returned from a query might be different for different servers.

Contributing

You can contribute in different ways

Report bugs

Report bugs at https://github.com/sjoerdk/clockify_api_client/issues.

Contribute code

Get the code

Fork this repo, create a feature branch

Set up environment

dicomtrolley uses poetry for dependency and package management

• Install poetry (see poetry docs)
• Create a virtual env. Go to the folder where cloned dicomtrolley and use

  poetry install 
  

• Install pre-commit hooks.

  pre-commit install
  

Add your code

Make your code contributions. Make sure document and add tests for new features. To automatically publish to pypi, increment the version number. See below.

Lint your code

• Run all tests
• Run pre-commit:

  pre-commit run
  

Publish

Create a pull request

Incrementing the version number

A merged pull request will only be published to pypi if it has a new version number. To bump dicomtrolley's version, do the following.

• dicomtrolley uses semantic versioning Check whether your addition is a PATCH, MINOR or MAJOR version.

• Manually increment the version number:

  • pyproject.toml -> version = "0.1.2"

• Add a brief description of your updates new version to HISTORY.md