Enhanced CT-MR Profile - Brief Proposal

1. Proposed Profile: Enhanced MR/CT Image

• Proposal Editor: Cor Loef/Dick Donker
• Date: 2007-08-23
• Version: 0.1
• Domain: Radiology

2. The Problem

The original DICOM CT, MR, XR, PET and US objects needed significant revision because of technology advancements, and new acquisition protocols. DICOM has published the specification for Enhanced Multi-frame MR (2003), CT (2004), XR (2006) and PET (2007). US is expected in 2008.

Products have started creating and storing enhanced CT and MR objects. This requires consistent hanging and display on workstations for the multitude of clinical acquisition protocols supported by these objects.

The encoding flexibility of these objects raises a real risk that modality vendors will select slightly different encodings for the same clinical use case, and consequently workstation vendors will be confronted with these variations. This may result in inconsistencies when hanging and displaying these objects. Users will be annoyed by unnecessary viewing differences, and interoperability is at risk.

A profile that specifies encoding of critical concepts for particular clinical use cases would greatly simplify the task of configuring hanging protocols during PACS and modality installation, and reduce the unnecessary and unwanted variance between sites.

3. Key Use Case

The Advanced CT/MR Taskforce has defined key clinical use cases for a demonstration at RSNA 2005:

• CT/MR CARDIAC IMAGING

New CT/MR cardiac images containing standard attributes for location and timing information.

• MR DIFFUSION

MR diffusion images containing standard attributes for diffusion B values, direction and diffusion image type.

• CT/MR CONTRAST PERFUSION

Intra-venous contrast enhanced images containing standard attributes for timing and phase.

• MULTI-STACK SPINE IMAGING

Images are acquired at multiple disk space levels of the spine such that multiple slices at each level are parallel to the disk space, each in a separate stack from slices at different angles through other disks. These images are exchanged in an interoperable way with standard attributes describing the stacks and their geometries.

• MR SPECTROSCOPY

Single-voxel, multi-voxel or multi-slice 1D MR spectra are exchanged with reference and metabolite images that contain information to allow correct display, post-processing and analysis.

4. Standards & Systems

The systems involved are Acquisition Modalities, PACS, and Workstations (Display, Processing, Reporting).

The standards involved are the DICOM 2006: Enhanced MR Image, MR Spectroscopy, Enhanced CT Image, Enhanced XA/XRF, Enhanced PET objects and possibly the current US Supplement.

5. Discussion

These use cases may constitute the basis for one or more new Image Content profiles, which specifies, as applicable:

• The required attributes (mostly already defined in the standard),
• The relevant functional grouping
• The content of the dimension module
• The use of a Real World Values LUT
• The use of a Supplemental Color LUT
• The relevant attributes for display on a workstation

A key decision will be how many of the new objects to address this year.

The NEMA Committee for the Advancement of DICOM would like to organize a first informal testing of this profile with systems that envisage or already give support for enhanced CT and MR objects at the January 2008 connectathon.

Issues needing Addressing

Downgrading to Old Objects

How should a PACS handle it if the modality provided Enhanced but a client needed old style. Creation of the old objects is not so bad, and could be triggered based on the SOPs in the association for the C-STORE. The problem is that before the C-STORE there is probably a C-FIND and when you don't know what the client wants, how many series do you tell them there is. One Enhanced, but if they want Old, there are many situations you would want/need to split into multiple series. Similar when they query for image instances.

Realistically, I think that this would have to be done on the basis of the requesting AET, keeping a table in the SCP as to which AETs want "old" and which want "new" CT objects (or similarly for MR of course). This could either be configured manually (yuk!), or perhaps better done by "trying" to negotiate the enhanced objects in each outgoing C-STORE operation...that way any upgrade of the client could automatically be updated as soon as t gets upgraded (well ONE query later in fact!) Once the decision has been made, then splitting into series/images is probably not too difficult......if anyone would like to send me an example enhanced image would would benefit from splitting, then I'd happily give it a try - Dave Harvey

Upgrading to New Objects

Taking old objects and generating enhanced is even harder since there are required fields that would need to be filled and the necessary information is likely not present.

I agree with previous suggestions that this is unlikely to be helpful or useful, and could set a dangerous precedent for bad image construction by primary producers who should be filling the data out properly!