NA-MIC Project Weeks

Back to Projects List

Linking segmentation and imaging data with different geometries using dcmqi

Key Investigators

• Reuben Dorent (BWH, USA)
• Steve Pieper (Isomics, Inc., USA)
• Tina Kapur (BWH, USA)
• Colton Barr (Queen's University, Canada)
• Andrey Fedorov (BWH, USA)
• Michael Onken (OpenConnections, Germany)
• David Clunie (PixelMed, USA)

Presenter location: In-person

Project Description

This project aims to clarify best practices and discuss possible fixes for linking segmentation and imaging data with different geometries using dcmqi. This question arose during curation of The Brain Resection Multimodal Imaging Database (ReMIND) for TCIA and subsequent challenges with viewing unlinked segmentations using the OHIF viewer.

Objective

1. Objective A: Discuss possible fixes for this dcmqi linking issue when using imaging data and segmentations with differing geometries.
2. Objective B: Ensure the ReMIND data available in TCIA has the appropriate linking between segmentations and imaging, and can be viewed using OHIF.

Approach and Plan

1. Get input from all stakeholders on best practices and potential fixes for this linking problem.
2. Establish a plan for next-steps.

Progress and Next Steps

1. Andrey summarized the implementation challenges in this issue - need to hear from @dclunie: https://github.com/QIICR/dcmqi/issues/489

2. Perhaps consider something like recent RT Structure Set addition to standard that added SourceSeriesInformationSequence and SourceSeriesSequence (CP 2296 Provide additional ROI parameters to avoid parsing strings)

3. Steve and David discussed that a more general solution could be to enable OHIF to render segmentation on top of any data that shares the same FrameOfReferenceUID. This would not require any change in DICOM or in the data and would benefit other usecases. This topic will be researched during the next IDC viewers coordinatation meeting.

Illustrations

No response

Background and References

No response