Dear Floris
Many thanks for your reply. You raise the issue of manufacturer support. We have been talking to all 3 major manufacturers with positive responses. For the PET data, we will likely get file format descriptions or access to converters to a standard file format. For MR, we will at least initially have to rely on existing research agreements and software to be able to convert to ISMRMRDF. If we need any extra lobbying, I'll let you know!
Support for PET list mode data would of course be desirable. It's also been requested by Jeff Fessler. As you've seen, our intention is make a "framework" that can use different underlying "engines" or "packages" for the actual work. As far as I know, the only open source package for PET reconstruction that supports list mode is STIR. I obviously have a strong interest in extending the list mode support in STIR. For instance, Siemens Biograph data will be supported in the forthcoming release of STIR, thanks to their public release of the file format. So, I would say that the CCP-PETMR software could support list mode data when using STIR. A minimum would be to create sinograms from the list mode data. Actual list mode reconstruction (partially supported by STIR) is lower priority for this CCP in my opinion, but feel free to disagree.
I will edit the document along these lines.
Kind regards
Kris Thielemans
Senior Lecturer at University College London,
Institute of Nuclear Medicine, UCL Hospital
235 Euston Road, London NW1 2BU, UK
-----Original Message-----
From: F.H.P.van_Velden
Sent: 12 August 2015 12:42
To: [log in to unmask]
Subject: RE: CCP PET-MR: request for input on software specification
Dear Kris,
I just saw your message on the Gate mailing list. I'm happy to hear that there will be open source software package that will enable research in image reconstruction for PET-MR. It seems that manufacturers are more and more protecting their codes and access to their reconstruction software, thereby unfortunately complicating research in the reconstruction field. So having an open source framework is definitely great news and of great value to the PET community!
I've read through the draft user specification document. One thing that came to mind is that you are initially planning to support projection data. In the future, will you also consider to support event data (list mode data) and list mode reconstructions? As far as I know, all Philips reconstructions are based on list mode files without histogramming them into sinograms prior to reconstruction. Supporting list mode files would definitely be beneficial for sites with Philips PET/MRs.
If I can think of other suggestions, I'll let you know. Enjoy your summer as well!
With kind regards,
Floris van Velden
-------------------------------
F.H.P. van Velden, PhD
Medical Physicist
Leiden University Medical Center (LUMC)
Nuclear Medicine, Radiology
PO Box 9600 zone C02-S
2300 RC Leiden
The Netherlands
www.lumc.nl
|