You are here : Home > Basis material decomposition method for material discrimination with a new spectrometric X-ray imaging detector

Publications

Basis material decomposition method for material discrimination with a new spectrometric X-ray imaging detector

Published on 29 March 2018
Basis material decomposition method for material discrimination with a new spectrometric X-ray imaging detector
Description
 
Date 
Authors
Brambilla A., Gorecki A., Potop A., Paulus C., Verger L.
Year2017-0412
Source-TitleJournal of Instrumentation
Affiliations
Univ. Grenoble Alpes, CEA-LETI, DTBS/LDET, Grenoble, France
Abstract
Energy sensitive photon counting X-ray detectors provide energy dependent information which can be exploited for material identification. The attenuation of an X-ray beam as a function of energy depends on the effective atomic number Zeff and the density. However, the measured attenuation is degraded by the imperfections of the detector response such as charge sharing or pile-up. These imperfections lead to non-linearities that limit the benefits of energy resolved imaging. This work aims to implement a basis material decomposition method which overcomes these problems. Basis material decomposition is based on the fact that the attenuation of any material or complex object can be accurately reproduced by a combination of equivalent thicknesses of basis materials. Our method is based on a calibration phase to learn the response of the detector for different combinations of thicknesses of the basis materials. The decomposition algorithm finds the thicknesses of basis material whose spectrum is closest to the measurement, using a maximum likelihood criterion assuming a Poisson law distribution of photon counts for each energy bin. The method was used with a ME100 linear array spectrometric X-ray imager to decompose different plastic materials on a Polyethylene and Polyvinyl Chloride base. The resulting equivalent thicknesses were used to estimate the effective atomic number Zeff. The results are in good agreement with the theoretical Zeff, regardless of the plastic sample thickness. The linear behaviour of the equivalent lengths makes it possible to process overlapped materials. Moreover, the method was tested with a 3 materials base by adding gadolinium, whose K-edge is not taken into account by the other two materials. The proposed method has the advantage that it can be used with any number of energy channels, taking full advantage of the high energy resolution of the ME100 detector. Although in principle two channels are sufficient, experimental measurements show that the use of a high number of channels significantly improves the accuracy of decomposition by reducing noise and systematic bias. © 2017 IOP Publishing Ltd and Sissa Medialab.
Author-Keywords
computer-aided diagnosis, Data processing methods, Detection of explosives, Medical-image reconstruction methods and algorithms
Index-Keywords
Atoms, Computer aided diagnosis, Data handling, Diagnosis, Explosives detection, Image reconstruction, Maximum likelihood, Photons, Piles, Poisson distribution, Polyvinyl chlorides, Spectrometry, X ray analysis, Data processing methods, Decomposition algorithm, Detection of explosives, Effective atomic number, Material decomposition, Material discrimination, Material identification, Maximum likelihood criteria, Medical imaging
ISSN17480221
LinkLink

Go back to list