 |
 |
 |
Poster communications
Component Separation for Spectral X-Ray Imaging Using the XPAD3 Hybrid Pixel Camera
Abstract : The advent of hybrid pixel cameras in X-ray imaging opens the way to the acquisition of spectral measurements. These new devices for which photon counting replaces charge integration incorporate a dedicated readout electronic for each pixel including a capability of selecting energies via the setup of an energy threshold. This ability is of uppermost importance for the development of new polychromatic X-ray imaging approaches that will exploit spectral information on the detected X-rays. Spectral measurements in X-ray imaging pave the way to the separation of images in several components of physical and biological interest: the photoelectric and the Compton contributions can be separated while several contrast agents can be simultaneously localized. We investigate the capability to perform component separation by using the newly developed XPAD3 hybrid pixel camera incorporated in the micro-CT demonstrator PIXSCAN. Firstly, we propose an approach to configure the acquisition setup in order to optimize the component separation problem with respect to the robustness to the photon noise. The method is based on the Cramer-Rao Bound (CRB) that indicates the lowest reachable variance for the estimation of each component whatever the algorithm. Secondly, we investigate the separation problem with two components namely the photoelectric and the Compton ones. We show on noisy simulated data that such a separation with optimized setup i) enhances the contrast and the Contrast to Noise Ratio (CNR) between biological materials (adipose, soft tissues) and water; ii) cancels the artifacts of the beam-hardening effect that may strongly degrade the image quality. On going work involves two steps: first, dealing with Monte Carlo simulations and real data acquired with the PIXSCAN demonstrator; second, dealing with component separation with more than two components by adding several contrast agents, for which PIXSCAN has already proved its ability to separate them.
Cited literature [6 references]
http://hal.in2p3.fr/in2p3-00956664
Contributor : Danielle Cristofol <>
Submitted on : Wednesday, April 23, 2014 - 5:39:55 PM
Last modification on : Thursday, January 23, 2020 - 6:22:14 PM
Long-term archiving on: : Sunday, April 9, 2017 - 10:20:17 PM
Contributor : Danielle Cristofol <>
Submitted on : Wednesday, April 23, 2014 - 5:39:55 PM
Last modification on : Thursday, January 23, 2020 - 6:22:14 PM
Long-term archiving on: : Sunday, April 9, 2017 - 10:20:17 PM
File
MDupont_IEEE_NSSMIC_13-1.pdf
Files produced by the author(s)