Edit this page

ProjectWeek

Back to Projects List

DICOM for Quantitative Imaging

Key Investigators

Project Description

This project aims to provide a central location for coordinating development of DICOM as applied to quantitative imaging applications and evaluation of implementation in various tools being developed at the Project week.

Objective

  1. Provide support (datasets, discussion of the standard, pointers to existing implementations, validation) to developers implementing DICOM Segmentation image, DICOM Parametric map storage and DICOM SR TID1500 “Imaging measurements” template.
  2. Document the status of support of those objects in tools participating in the project week.
  3. Discuss use cases.
  4. Help researchers aiming to use DICOM for their specific use cases.

Approach and Plan

  1. Add pointers to the relevant existing resources.
  2. Hold discussions with the developers.
  3. Perform evaluation of the new implementations, assemble sample datasets generated by the implementations. Document the status. Projects implementing relevant functionality: OHIF, XNAT, MeVisLab, pydicom, DCM4CHEE, MITK …
  4. Anyone interested in learning how DICOM can be used to help organizing their research data can bring a dataset to discuss and evaluate support in tools as applied to their use case.

Progress and Next Steps

DICOM Breakout

Several relevant projects were presented during the DICOM breakout session on Wednesday:

Resources:

DICOM SR TID1500 support in pydicom

Markus has been working on defining/implementing the API, helped by the discussions/pointers from Andrey and Steve

Infrastructure improvements

  1. Until recently, MITK was using a patched version of dcmqi for DICOM SEG/SR1500 support due to the issues in dcmqi superbuild that prevented its use without SlicerExecutionModel. This was resolved (jsoncpp updated to a version more suitable for C++11) with the joint efforts of JC, Andrey and Marco, and now we can expect MITK to use the main dcmqi repository.
  2. JC fixed packaging of dcmqi on Linux, and improved the CI script for CircleCI. Thanks for the magic, JC!

New tools supporting DICOM for derived data

  1. Srikrishna is working to use DICOM SEG for communicating analysis results within the Siemens Teamplay project. Functional demo is available. Sample dataset to be shared to DICOM4QI.
  2. James has a functional implementation of DICOM SEG read/write in XNAT, scheduled to become available in the next release in April 2019.
  1. Converting non-DICOM images into DICOM Raised by Tobias Stein and Maro: In imaging research groups, there is frequently an abundance of the images in non-DICOM format. There is a practical question of how to harmonize those with the DICOM data, and maintain them in the same database. It would be helpful to be able to convert those representations into some DICOM format. Slicer has a CreateDICOMSeries module, but it can only create a non-enhanced CT series, and has limitations. It will also be, in the general case, impossible to meaningfully populate modality-specific attributes, if a modality-specific object is created. Discussed possible suitable containers for such representation:
    • Raw Data IOD - does not have PixelData, may be “too raw”?
    • Various types of Secondary Capture IODs
    • Does it make sense to suggest a new IOD to cover an image of a known modality, but where acquisition details are not known?
  2. Harmonization of DICOM infrastructure between Slicer and MITK (or rather, lack of such!). Both MITK and Slicer (need to) implement various strategies for parsing DICOM data into volumetric reconstructions. Both do this independently, as many other tools and converters do. It might be interesting to investigate how those developments could be coordinated. Andras, Marco and Andrey discussed this and are planning to follow up with a further discussion to go over details. Will also invite Ralf Floca from the MITK team.

Illustrations

Background and References