Search

International audience

International audience; Defect perception on films by operators is difficult to compare to the sameperception on simulated images due to the fact that operators are not used to visualizedigital images, and moreover, the conditions of visualization have a great influenceon the perception. In order to get round of the problem, we have compared the operatordecision on radiographic films with quantitative parameters computed on thecorresponding simulated images. The aim is to define detectability parameters whichhave the best correlation with the operator decision. In order to get statisticalsignificance, a high number of “defects” has been used by using image qualityindicators which were cut in several parts. The simulation of the complete scene wasa real challenge due to the high resolution needed combined with the quite large sizeof the steel plate. We present the correlation results obtained with differentparameters, where it appears that all parameters based on contrast correlate very wellto the operators’ decision. As a second step, a simulation study was conducted in orderto investigate the influence of a density gradient, which is known to decrease thevisibility of defects. A global detectability criterion is finally proposed.

International audience; Computed Radiography is a potential replacement technology fortraditional film radiography, promising shorter exposure times and a wider exposurerange while maintaining the advantage of detector flexibility, which is relevant forinstance for panoramic exposures.Computer modeling can play a supporting role in the handling of the transitionto a new inspection technique, facilitating the definition of the optimal operatingconditions. We present the result of a thesis carried out at INSA Lyon and supportedby Areva, EDF and GE, aiming to develop a computer model for the entire acquisitionchain from exposure to optical read-out of an imaging plate, as well as its actualimplementation in an industrial code, and an illustration of its application.

International audience; Computed radiography (CR) is a digital radiographic technique based on the use of a phosphor imaging plate (IP). In CR, X-ray flux is received by IP forming a latent image which is later read and digitized by an optical system with a scanning laser. CR is an effective replacement of the traditional film-based radiography in medical applications for its high dynamic range and high efficiency at low energies. For industrial applications, it cannot meet all inspection requirements because of its poor performance at high energies. In terms of spectral resolution, the IP is strongly energy dependent and much more sensitive to energies up to about 100keV. In terms of spatial resolution, the Point Spread Function of the IP is also degraded at high energies. Moreover the optical readout process further degrades the final image quality. To be fully used in industrial radiography, CR performance needs to be optimized. Guidelines have been addressed in international standards to ensure a good image quality supplied by CR system, where source type and metallic intensifying screens are defined for different cases. Adjusting the scanning parameters might also be a way to improve the final image quality. Computer simulation can help to investigate underlying problems and improve the system performance. A lot of prior studies focused on the CR exposure, while this study focuses more on the simulation of the CR readout process including the laser scattering within the IP and the laser scanning effect. The talk will first present briefly the CR simulation approach we developed, and then detail the optical readout simulation. A Monte Carlo method is adopted for simulating light transport in the IP. The influence of several parameters such as physical properties of IP material and IP thickness are studied. During the laser scanning process, the latent image is modified as the CR readout is a destructive process. The laser scanning process is simulated by an analytical model taking into account the scanning parameters like laser power or speed displacement. The influence of those parameters on final image quality will be presented.

International audience; Computed tomography (CT) is a relevant nondestructive inspection method for composite material. However, X-ray attenuation contrast difference between fibers and matrix can be low, which makes identification of defects difficult. Optimization of the tomographic process has then to be considered, such as studying the acquisition parameters and phenomena that decrease the contrast. Simulation of the CT method is considered in order to optimize contrast between fibers and matrix. Simulation allows to study phenomena independently to the others. Phantom was built in order to check the accuracy of the computation. Even though results show discrepancies between computation and experimental measurements, these discrepancies have been found to be due to an experimental artefact. When corrected, simulation fits very well with experimental results and this procedure allows to seek an optimal setup for composite material inspection.

International audience; Dual energy X-ray transmission technique is used since many years in the medical field for bone /soft tissues separation, as well as in luggage inspection for light/heavy materials distinction. Its application to waste sorting first implies a feasibility study for performance assessment of different sorting scenarios. We have used the “Virtual X-ray Imaging” simulation software developed at INSA-Lyon (LVA laboratory) in order to optimize the dual line detector geometry (sandwich or parallel), but also to assess performances of the method.

International audience; Computed Radiography (CR) based on photostimulable imaging plate (IP) is increasingly used in the field of non-destructive testing. For the inspections of high attenuation specimens employing high energy sources, the CR performance is poor because of the response of the IP. Simulation is a very useful tool to predict experimental outcomes and determine the optimal operating conditions. We propose a hybrid simulation approach which combines the use of both deterministic and MC codes for simulating rapidly a complex geometry imaging setup. This approach allows us to take into account the degradation effect introduced by X-ray scattering and fluorescence transport as well as the optical photon scattering taking place in IP during X-ray exposure and optical readout processes. The results of different simulation configurations are compared.

National audience; Dans ce papier, nous proposons une méthode de restauration de spectre acquis en en diffraction de rayon X en dispersion d'énergie (EDXRD). En utilisant un modèle physique de formation des données, nous suivons une approche maximum a posteriori pour réduire le flou causé par la mauvaise résolution spectrale du détecteur et par la collimation très grossière des systèmes EDXRD utilisant des tubes à rayon X. Cette méthode extrait d'un fond continu les pics causés par la diffraction par des matériaux cristallins. Testé sur des données réelles (diffraction d'un morceau de graphite), notre algorithme a permis de d'extraire les pics de diffraction avec une bonne précision (environs 1 keV), y compris à haute énergie où très peu de photons sont mesurés.

International audience; Computer modelling of non-destructive testing methods has come a long way from the beginnings in the mid 90s to today. In this article we review radiographic modelling from the perspective of two important application domains with very different requirements, nuclear and aerospace. Radiographic modelling for components with higher wall thicknesses, as they are typical for nuclear applications, must include precise predictions of scattered radiation and its impact in terms of contrast reduction. Dedicated or general purpose Monte Carlo methods with the ability to calculate higher order scattering events are the state of the art for these applications. The challenge here is to obtain the simulation results in reasonable computation time, and a number of acceleration schemes have been devised to that end. Aerospace applications, on the other hand, have stronger requirements on the modelling code's capabilities to import complex CAD geometries, and can benefit from faster analytical scatter models, limited to first or second order scattering events. Similar distinctions can be made for the various approaches proposed to accurately model geometrical and film unsharpness, film granularity, film responses, film/foil cartridges and photon noise. With the advent of digital radiography, computer modelling faces new challenges to take physical effects into account which could be neglected with traditional silver films.

International audience; The aim of the "SPIDERS" project is the validation of a novel portable imaging system for the characterization of dangerous/illicit materials inside objects. The system combines two X-ray techniques: the dual-energy tomosynthesis and the energy-dispersive diffraction. The latter has been chosen for its utility in the detection of narcotics and a wide range of explosives as pointed out elsewhere. The main goal is the 3D object exploration without imposing any movements to the object. In the diffraction modality, measurements are performed by translating the diffracted beam collimator and the detector. The latter is based on CdZnTe semiconductors, which can provide energy resolution (DE/E) of about 3% @ 90 keV and room temperature. We report here the optimization study for the system specifications in the diffraction modality. In order to select the experimental conditions (the kV of X-ray sources, the "diffraction" angle q, the angular resolution Dq/q, the collimator design, ...) best suited for the system purposes, a multi-parametric simulation study has been performed.

International audience; Scattered radiation is a fact of life in industrial radiography, and as such must also be taken care of in computer models for industrial radiography. The French national research project RADIOLA"aims to integrate the key components of three existing computer codes (VXI: Virtual X-ray Imaging, Sindbad and Moderato) into the CIVA platform and provides several means to calculate scatter predictions, depending on the configuration at hand. For high wall thicknesses, such as those typically encountered in nuclear applications, a traditional Monte Carlo scatter model is available, which provides reliable results at the expense of computation time. For components with smaller wall thicknesses, where first order scattering is dominant, an analytical approach is available, which is particularly well suited for aerospace applications. We present and illustrate both approaches, compare their respective application domains, and conclude with a proposal for a hybrid model which attempts to provide n-th order scatter predictions, which could be a computationally effective approach for intermediate wall thickness ranges.

International audience; Quantitative X-ray fluorescence measurements have been conducted on naturally lead-contaminated samples. The calibration procedure using the ratio of fluorescence to Compton scattered radiation was investigated using Monte Carlo simulation. Experimental results with low-energy photons (14 keV) and simulations show a very good linearity of the fluorescence to Compton ratio as a function of metal concentration. Lead (Pb), iron (Fe) and zinc (Zn) are measured in samples of "Phaseolus vulgaris" (bean seeds) which have been grown using a nutritive solution with different Pb doping. Natural contaminated samples are thus obtained. The calibration must be done for fixed conditions of X-ray energy and scattering angle, while X-ray beam intensity and detector to sample distance can change from one sample to another. Simulation allows to evaluate the matrix effect on the calibration curve, and shows that linearity is preserved even in the presence of other heavy elements in the fluorescence spectrum. However, calibration must be done using samples with similar matrix as it affects the slope of the curve.