Material decomposition and tomographic reconstruction for spectral photon counting computed tomography

Scientific context

The CREATIS laboratory (http://www.creatis.insa-lyon.fr) (Lyon, France) opens a postdoctoral research fellowship funded by the European H2020 project SPCCT (http://www.spcct.eu) and the French National Cancer Institute project SPEDIV. The objective of this postdoc is to develop and validate material decomposition and tomographic reconstruction for a prototype Spectral Photon Counting Computed Tomography (SPCCT) scanner and to compare it with single photon emission computed tomography (SPECT) quantification of gadolinium nanoparticles.

Dual-energy x-ray CT is becoming a widespread clinical tool thanks to the recent development of commercial scanners. K-edge imaging is going further using spectral photon counting detectors [1] to reconstruct the spatial distribution of a K-edge contrast material [2]. The prototype pre-clinical scanner reconstructs the K-edge image in a two-step process: (1) decomposition of the acquired data into a K-edge sinogram and two other materials, and (2) filtered backprojection (FBP) reconstruction. Promising results have been obtained but the sensitivity still needs to be increased.

Goal and tasks

A previous postdoc has investigated a better modeling of the statistics in an iterative tomographic reconstruction instead of the second FBP step [3] and one-step methods where the two steps are solved simultaneously [4]. The most promising algorithm is being implemented in the Reconstruction Toolkit (http://www.openrtk.org/). The postdoc will continue these developments to improve the performance in terms of accuracy (e.g. by adding pile-up to the forward model) and/or speed (e.g. by improving the GPU implementation). The developments will be validated on simulations and on the new prototype which will be installed before summer 2018, in collaboration with the biologists and the physicians of the projects. The comparison with SPECT will be a joined effort with another postdoc of the group.

Profile

  • Education: The candidate must hold a PhD in medical imaging, applied mathematics or physics.
  • Experience: Inverse problems, tomographic reconstruction, x-ray imaging.
  • Programming skills: C/C++, Python, CUDA.
  • Languages: English required, French optional. The candidate must have good communication skills and be experienced in scientific writing.
  • Location: Lyon, France.
  • Salary (net): depending on experience starting at 2000 euros / month.
  • Period: 1 year renewable, starting as soon as possible.

References

  1. K. Taguchi and J. Iwanczyk, “Vision 20/20: Single photon counting x-ray detectors in medical imaging,” Med Phys, vol. 40, no. 10, p. 100901, Oct 2013. [Online]. Available: http://dx.doi.org/10.1118/1.4820371

  2. J. Schlomka, E. Roessl, R. Dorscheid, S. Dill, G. Martens, T. Istel, C. Ba ̈umer, C. Herrmann, R. Steadman, G. Zeitler, A. Livne, and R. Proksa, “Experimental feasibility of multi-energy photon-counting K-edge imaging in pre-clinical computed tomography,” Phys Med Biol, vol. 53, no. 15, pp. 4031–4047, Aug 2008. [Online]. Available: http://dx.doi.org/10.1088/0031- 9155/53/15/002

  3. C. Mory, B. Brendel, K. Erhard, and S. Rit, “Generalized least squares for spectral and dual energy CT: a simulation study,” in Proc. of the fifth international meeting on image formation in X-ray CT, 2018. [Online]. Available: http://www.creatis.insa-lyon.fr/∼srit/biblio/mory2018.pdf

  4. C. Mory, B. Sixou, S. Si-Mohamed, L. Boussel, and S. Rit, “Comparison of five one-step reconstruction algorithms for spectral CT,” IEEE Transactions on Medical Imaging, Submitted. [Online]. Available: https://hal.archives- ouvertes.fr/hal- 01760845