Emil Sidky just sent me the following:Hi Igor,
I would like to point out recent work we did in collaboration with Hopkins on sparse-view CT in Physics of Medicine and Biology:
It's linked also on Jeff's I-STAR page (a lot of nice stuff there on CT image science):
It's a featured article, and therefore free access to anyone. It even got a little press on medicalphysicsweb.org
The main points of interest:
- The reconstructed images are evaluated objectively by a number of image quality metrics.
- The next point echo's Mark Neifeld's observation that all digital imaging has always been compressive. We show that even with the standard number of CT views, constrained TV-minimization will, in general, lead to a different image than standard filtered back-projection algorithms. (I always feel uneasy when I see the claim that CS beats the Nyquist sampling theorem, because truly fully-sampled data in CT does not exist.)
- Nonetheless, the CS-motivated idea of employing constrained, TV-minimization does appear to be robust. One criticism we often encounter is that such algorithms apply strictly to piece-wise constant images. Well, due to non-ideality of CT data relative to the X-ray transform image model, reconstructed images are not piecewise constant. Nevertheless, constrained, TV-minimization still yields high image quality for sparse-view CT data.
Jong Chul Ye also sent me this:
You might be interested in our recent paper which will appear in IEEE Trans. on Medical Imaging.Finally, Ronan Le Boulch sent me the following:
K. Lee, S. Tak, and J. C. Ye, "A data-driven sparse GLM for fMRI analysis using sparse dictionary learning with MDL criterion," IEEE Trans. Medical Imaging, Special issue on "Compressive sensing for biomedical imaging", November 2010, Accepted .
This paper demonstrates that compressive sensing theory is very useful for fMRI signal analysis by enabling a fully data-driven approach.
I am a french engineer working on the Metiss team with Remi Gribonval and I am currently updating the MPTK Pursuit Toolkit. I have updated MPTK Library to a 0.5.9 version since March of this year and I would like to know if it's possible to add a note or an annoucement of this release under the Nuit Blanche website.
I don't know if you know this library or if you have ever used it but here is a summary :
The Matching Pursuit Tool Kit (MPTK) provides a fast implementation of the Matching Pursuit algorithm for the sparse decomposition of multichannel signals. It comprises a C++ library, standalone command line utilities, and some scripts for running it and plotting the results through Matlab.
Some more informations can be found here : http://mptk.irisa.fr/
The downloads can be found here : https://gforge.inria.fr/frs/?group_id=36
The tracker for Bugs or Support requests can be found here : https://gforge.inria.fr/tracker/?group_id=36
Thanks Ronan, Jong and Emil for providing some context and by the same token making the web a better place. I'll feature your papers and libraries tomorrow.
Image: NASA/JPL/University of Arizona, Blocks in the Olympus Mons (PSP_003450_1975)
No comments:
Post a Comment