NA-MIC Project Weeks

Back to Projects List

Importing and displaying DICOM Structured Reports in Slicer

Key Investigators

• Deepa Krishnaswamy (Brigham and Women's Hospital, USA)
• Steve Pieper (Isomics, USA)
• Andras Lasso (Queen's University, Canada)
• Chris Bridge (Massachusetts General Hospital, USA)
• Ron Kikinis (Brigham and Women's Hospital, USA)
• Andrey Fedorov (Brigham and Women's Hospital, USA)

Project Description

In TCIA, multiple collections include CSV files with different measurements. For instance, the ProstateX collection has information about the Gleason score and target biopsy points. Other annotations, such as the lesion bounding boxes for NLST from Sybil are in json format.

There is no easy way to use these measurements. If we want to use them in Imaging Data Commons (IDC), the data needs to be in a standardized format like DICOM Structured Reports (SRs). From the last project week here, we have already created SRs for a few of these collections that hold points and bounding boxes.

However, right now, Slicer can only load a specific type of SR. We are currently working on adding functionality to load SRs to load and display points, boxes, and lines.

Objective

We will create SRs to hold points and bounding boxes, and modify the DICOMTID1500Plugin to load and display these measurements as markups.

Approach and Plan

1. Create SRs for the ProstateX (csv from TCIA) and NLST (Sybil) collections.
2. Modify the DICOMTID1500Plugin.py in QuantitativeReporting to read these SRs and display as markups.
3. Show a table holding the measurements.
4. Display the markups appropriately in the subject hierarchy.

Progress and Next Steps

1. We have created SRs for ProstateX (points) and NLST Sybil (bounding boxes).
2. We have modified the plugin to load points, boxes, and lines.
3. We have loaded the markups in a folder in the subject hierarchy and have displayed the table.
4. Currently, we are working on loading the correct referenced series.

I had some good discussions this week with Ron, Steve, and Andras about how to improve the extension.

Summary of the discussion:

• How to effectively display the different annotations
• How to store data describing each of the annotations - node attributes vs table
• Improvements in the subject hierarchy - making sure table/markups are linked to the series/study
• How this interface can be used to later save SRs
• How this work can fit into the larger scheme of data exploration
• How to modify and preconfigure the layout for each of the types of annotations, for instance, below for bounding boxes:
• 

Illustrations

Example of a DICOM SR with points - ProstateX biopsy target points

Example of a DICOM SR with bounding boxes - Sybil lesion annotations for NLST

Example of a DICOM SR with lines

Background and References

• Slicer discourse discussion
• My fork of QuantitativeReporting
• Creating SRs from last project week