Quantitative reconstruction of PIXE-tomography data for thin samples using GUPIX X-ray emission yields

Claire Michelet (1) , Philippe Barberet (1) , Guillaume Devès (1) , B. Bouguelmouna (1) , S. Bourret (1) , Marie-Hélène Delville (2) , Quentin Le Trequesser (2, 1) , N. Gordillo , D. G. Beasley (3, 4) , A. C. Marques (3) , R. Farau (1) , B. Toko (1) , J. Campbell (5) , J. Maxwell (5) , Philippe Moretto (1) , Hervé Seznec (1)

1 CENBG - Centre d'Etudes Nucléaires de Bordeaux Gradignan
2 ICMCB - Institut de Chimie de la Matière Condensée de Bordeaux
3 Centro de Ciências e Tecnologias Nucleares
4 CMIC - Centre for Medical Image Computing
5 Guelph-Waterloo Physics Institute

We present here a new development of the TomoRebuild software package, to perform quantitative Particle Induced X-ray Emission Tomography (PIXET) reconstruction. X-ray yields are obtained from the GUPIX code. The GUPIX data base is available for protons up to 5 MeV and also in the 20–100 MeV energy range, deuterons up to 6 MeV, 3He and alphas up to 12 MeV. In this version, X-ray yields are calculated for thin samples, i.e. without simulating X-ray attenuation. PIXET data reconstruction is kept as long as possible independent from Scanning Transmission Ion Microscopy Tomography (STIMT). In this way, the local mass distribution (in g/cm3) of each X-ray emitting element is reconstructed in all voxels of the analyzed volume, only from PIXET data, without the need of associated STIMT data. Only the very last step of data analysis requires STIMT data, in order to normalize PIXET data to obtain concentration distributions, in terms of normalized mass fractions (in μg/g). For this, a noise correction procedure has been designed in ImageJ. Moreover sinogram or image misalignment can be corrected, as well as the difference in beam size between the two experiments. The main features of the TomoRebuild code, user friendly design and modular C++ implementation, were kept. The software package is portable and can run on Windows and Linux operating systems. An optional user-friendly graphic interface was designed in Java, as a plugin for the ImageJ graphic software package. Reconstruction examples are presented from biological specimens of Caenorhabditis elegans – a small nematode constituting a reference model for biology studies. The reconstruction results are compared between the different codes TomoRebuild, DISRA and JPIXET, and different reconstruction methods: Filtered BackProjection (FBP) and Maximum Likelihood Expectation Maximization (MLEM).

Mots clés: PIXE tomography; MLEM; Filtered backprojection; ImageJ; Caenorhabditis elegans

Domaines: Matériaux

Claire Michelet, Philippe Barberet, Guillaume Devès, B. Bouguelmouna, S. Bourret, et al.. Quantitative reconstruction of PIXE-tomography data for thin samples using GUPIX X-ray emission yields. Nuclear Instruments and Methods in Physics Research Section B: Beam Interactions with Materials and Atoms, 2015, 348, pp.92-99. ⟨10.1016/j.nimb.2014.11.035⟩. ⟨hal-01144075⟩