2014 Summer Project Week: CLI Derived DICOM Data

• 

  Projects List

Key Investigators

• Steve Pieper, Isomics
• Andrey Fedorov, BWH
• Jim Miller, GE
• David Clunie, PixelMed (available to answer questions)
• Michael Onken, Open Connections (providing input remotely)

Project Description

For QIICR we want to represent all data in DICOM for three specific clinical trials workflows (prostate MR DCE, GBM MR DCT, H&N PETCT). To make Slicer compatible with this goal, we wish to have a framework to create correct DICOM files for the results of CLI and interactive processing.

Objective

• See if it's possible to create a wrapper that handles the DICOM I/O for non-DICOM-aware CLIs
• See what extra information from CLIs would be required to know how to wrap them
• See if it makes more sense to have CLIs work more natively with DICOM rather than wrapping them

Approach, Plan

• Review some sample derived images generated by other code or commercial vendors (David to validate they are correct)
• Develop a reference implementation of DCMTK C++ code that creates derived objects for a simple workflow like denoising of an MR storage object. (Michael to provide code, David to validate correctness)
• Review specific CLIs in Slicer to see what information they need as input and how the output should be represented (Steve and Andrey)
  • DCE processing
  • image arithmetic including resampling (i.e. images of different orientations)
• Look at how Slicer Execution Model might express the information needed to work with this data (Jim, Andrey, Steve)

Progress

• Held DICOM breakout session and discussed use cases for derived data
• Discussed options for holding data in Subject Hierarchy vs node attributes (prefer Subject Hierarchy wherever possible)
• Michael will deliver DCMTK sample code by the end of June